The present study was conducted between June 2018 and June 2022 and the honeys collected in several localities. Some of the honeys were collected in the Mpassa and Sébé-Bricolo departments of the Haut-Ogooué province in south eastern of Gabon, the other honey was obtained from Komo-Mondah department of Estuaire province ( Figure 1 ). The Mpassa department is located in the central part of the province, with the town of Franceville as its capital. The village Moulendé (S 1˚37'74'', E 13˚27'18.46'') on the Franceville-Moanda axis ( Figure 2 ), around 15 km from Franceville, was selected as the experimental site

for two of the five honey samples. The Moulendé and Benguia areas, in the Mpassa Department, are characterized by vast expanses of grassy and shrubby savannahs surrounding smaller areas of forest known as savannah-gallery forest complexes or forest-savannah mosaics [12] (Demichelis et al., 2022), [13] (Maloba Makanga, 2011), [14] (Mikissa et al., 2008), [15] (Orijemie, 1990).

The beekeeping in the experimental apiary in Moulendé was carried out using Kenyan-type hives with mobile bars. The priming to attract the bees was done by inducing a bait consisting of honey and inflorescences of Ocimum gratissimum on each bar and on the internal walls of the hive. The honeys were harvested in the traditional way but with care in handling. Honey extraction was carried out by manual pressing of the wax comb. The honey obtained was filtered using strainers with various orifices of 2 mm and 0.5 mm, respectively.

Five honey samples (M1 to M5) were collected in 2019 and 2022 in several localities. Honey samples M1 and M3 come directly from the experimental apiary in a savannah-gallery forest complex of Moulendé (Haut-Ogooué province).

The honey sample M4 was purchased from a beekeeper in the savannah-gallery forest Complex of Benguia (Haut-Ogooué province). The honey samples M2 and M5 were respectively purchased from beekeepers in Okondja primary forest (Haut-Ogooué province) and in Libreville primary forest (Estuaire province). In Table 1 , honeys are classified according to their ecosystem (savannah-gallery forest complex and Primary forest) and origin. Glass bottles (250 ml) for samples M1 and M3 were first sterilized with 3% diluted bleach (NaClO), rinsed several times with distilled water and oven-dried at 105˚C for 24 hours. After sampling, the bottles were sealed and stored in a dark place at room temperature. The packaging and storage methods used for M2, M4 and M5 honeys are not known. The M4 and M5 honeys were purchased in 1L glass bottles, while the M2 honey was in a 500 ml plastic container. As soon as purchased, they were transferred into 250 ml sterilized glass bottles and stored in a dark place at room temperature.

M1 to M5: Various honeys.

Physico-chemical parameters were analyzed using the official methods of analysis of the Association of Official Analytical Chemists [16] (AOAC, 1990) and the harmonized methods of the International Honey Commission [17] (Bogdanov, 1999). Samples were analyzed over the same period to ensure uniformity of conditions and comparability.

Hydrogen potential (pH) was measured according to the method described by [16] (AOAC, 1990) using a pH meter (Consort C860 (USA), with an accuracy of ± 0.002 pH units). To measure the pH of honey samples diluted to 10% (w/v), 10 g of honey were diluted in 75 ml of distilled water before the probe was immersed.

Free acidity was measured according to the method described by [2] (Codex Alimentarius, 2022) using a 10% honey solution. In the 10% honey solution used to determine the pH of each sample, NaOH (0.1 N) was added dropwise using a burette, until a pH of 8.3 was obtained. The volume of NaOH used was then reported. Free acidity, expressed in milli equivalents of free acid per kilogram (meq∙kg⁻¹), is calculated according to equation 1 [18] (MAFF, 1992).

$\text{AL}=\frac{1000\times \text{V}\times \text{N}}{\text{m}}$ (1)

Where AL: Free Acidity (meq∙kg⁻¹), V: volume (mL) of NaOH used to neutralize honey to 10%, N: normality of NaOH used and m: mass of honey (g) actually used for measurement.

Water content was determined using an Abbé refractometer (ZEISS refractometer, West Germany). All measurements were taken at 28˚C and corrected to 20˚C. A drop of honey is placed on the refractometer. The water content of the honey sample was determined according to Equation (2) [19] (Dailly, 2008).

$\text{Water content}\left(\text{\%}\right)=\frac{\left[-0,2681-\log \left(n_D^{20}-1\right)\right]}{0,002243}$ (2)

Where $n_D^{20}$ is the Abbé refractive index determined at 20˚C.

The electrical conductivity (EC) of a 10% solution of each type of honey was measured using a WTW-LF 330 conductivity meter (Bioblock, Germany). The technique is based on measuring the electrical resistance at 20˚C in a 10% honey solution with 10 g honey in 75 ml distilled water [6] (Belhaj et al., 2015), [16] (AOAC, 1990). Results were expressed in milli Siemens per centimeter (mS∙cm⁻¹) according to the Codex Alimentarius standard.

To determine honey density, 5 ml of distilled water were weighed, followed by 5 ml of honey on a Highland® Portable Precision Balances-HCB 153. Density is expressed according to Equation (3).

$\text{density}=\frac{mhoney}{mwater}$ (3)

The total sugar content of each sample was obtained by analyzing a drop of honey using a portable Brix refractometer, type ATC, at 20˚C with direct reading. The result is expressed in Brix (equivalent to 1 g sucrose per 100 g solution).

Data were obtained from three replicates of each honey sample and results expressed as means ± mean standard deviation. Results were processed using XLSTAT 2022 software. An analysis of variance (ANOVA) was performed at the 5% probability threshold between the different honeys for the physico-chemical parameters to assess possible differences. The Pearson correlation test with 95% confidence interval was used to determine the correlation between the different physico-chemical parameters of the honeys studied, and the covariance matrix was subjected to principal component analysis (PCA).

The pH of honey samples is crucial during the extraction process, as it affects the stability, texture and life expectancy of different honeys. The pH is sufficiently acidic to slow or inhibit the growth of many species of bacteria [5] (Abselami et al., 2018), [20] (Naman et al., 2005). The values obtained in Table 2 show average pH values for the different honeys collected ranging from 3.4 ± 0.10 (M5) to 3.9 ± 0.17 (M1). A significant difference (p < 0.05) is observed between M1 and M4 honeys. These values (pH between 3.5 and 4.5) are in line with the recommendations of [2] (Codex Alimentarius, 2022). These pH results are similar to those reported for other honey samples originating from Burkina-Faso, Brazil, Morocco, and Algeria [5] (Abselami et al., 2018) and [21] (Backchiche et al., 2018), [22] (Meda et al., 2005). The conformity of these pH values gives an indication of the freshness of the honey and helps to confirm the botanical origin (nectar honey), both of honeys from the savannah-gallery forest complex (M1, M3 and M4) and those from the primary forest ecosystem (M2 and M5). However, [7] (Mbogning et al., 2011) found a pH value higher than both our results and those of [2] (Codex Alimentarius, 2022) and [23] (Conseil de l’Union Européenne, 2002)) for honey harvested during the rainy season in north-west Cameroon. Similarly, [3] (Koudegnan et al., 2012) found higher values for some sustainably preserved honeys with an average pH of 5.11 than in the present study. Honey is naturally acidic, regardless of its geographical origin. This acidic pH is due to the presence of various organic acids (gluconic acid, acetic acid and citric acid) contained in the nectar [24] (Khalil et al., 2012), but above all to the bee’s salivary secretions and enzymatic processes [20] (Naman et al., 2005). Organic acids contribute to honey’s flavor and stability against microbial deterioration. As a result, whatever the ecosystem - savannah-gallery forest or primary forest the pH of our honey samples is naturally acidic, in compliance with [2] (Codex Alimentarius, 2022).

Honey acidity is an important parameter for assessing its quality. It gives very important indications of its condition [25] (Bogdanov, 1997). In fact, the determination of free acidity makes it possible to determine the possible alteration of honey over time. The values obtained from the analysis of the five (5) honeys ( Table 2 ) show that average acidity values vary between 37.12 meq∙kg⁻¹ for honey M3 (Moulendé savannah-gallery forest) and 76.65 meq∙kg⁻¹ for honey M2 (Libreville forest). A significant difference (p < 0.05) was observed between M3, M5 and M1, M2 and M4 honeys. The values for M3 (Moulendé savannah-gallery forest) and M5 (Okondja forest) are in line with the [2] (Codex Alimentarius, 2022) standard of 50 meq∙kg⁻¹, while the acidity values for M1 (71.54 meq∙kg⁻¹), M2 (76.65 meq∙kg⁻¹) and M4 (75.50 meq∙kg⁻¹) are significantly higher. Regarding our values ( Table 2 ), the variation in free acidity does not seem to depend on the type of complex savannah-gallery forest or primary forest ecosystem. According to [26] (Pérez-Arquillué et al., 1995), this variation in free acidity can be attributed to the floral origin of the honey [27] (Schweitzer, 2004), on the other hand, shows that the natural acidity of honey increases with the stage of ripening (aging), whether it contains propolis or is altered by fermentation. These results are in line with the values obtained by [10] (Djossou et al., 2013), which range from 36.92 and 64.38 meq∙kg⁻¹ for honeys marketed in Benin. The values obtained for honeys M3, M4 and M5 are in agreement with the value of multifloral honey from Morocco obtained by [6] (Belhaj et al., 2015).

Determining the water content of honey is very important for its preservation during storage. The water content of honey can be favorable either to undesirable fermentation of this honey due to the action of osmo-tolerant yeasts, or to hygroscopy of the honey [4] (Machado De-Melo et al., 2018). For example, when the water content of honey is below 18% and air humidity is above 60%, honey absorbs moisture from the air. Moisture content also depends on the honey's stage of ripening at the time of harvest. The honey’s state of ripeness depends on the moment chosen between the start of the ventilation and nectar trophallaxis process by the bees and the harvesting of the honey [5] (Abselami et al., 2018), [20] (Naman et al., 2005), [25] (Bogdanov, 1997), [28] (Gomes et al., 2010). Water content varied between honey samples, ranging from 17.49 ± 1.25 to 21.21 ± 1.25 ( Table 2 ). A significant difference (p < 0.05) was observed between honey samples M3, M4 and M5. Honey M4 has the lowest water content (17.49% ± 1.25%), while M3 has the highest (21.21% ± 1.25%). According to [2] (Codex Alimentarius, 2022) on honey quality, water content must be less than 21%. The values of the honey samples studied comply with this Codex requirement irrespective of the different ecosystems. The significant differences between the water contents of all the honeys can be explained by the large number of factors influencing the moisture content of honey mentioned above. Numerous authors have obtained water content values for natural multi-floral honeys similar to those in the present study and in compliance with the Codex Alimentarius standard [5] (Abselami et al., 2018), [10] (Djossou et al., 2013), [29] (Albu et al., 2021) [30] , (Krishnasree and Ukkuru, 2017). The lowest water content values are comparable to those found by [10] (Djossou et al., 2013) in their work on honeys marketed in Benin (9.84% - 19.76%) and the highest values comparable to those of [31] (Buveka Ngoma et al., 2018) on natural honeys in the dry season in the Democratic Republic of Congo (20% - 21%).

Electrical conductivity expresses the ability of an aqueous solution to conduct an electric current. It is positively correlated with soluble salt content [6] (Belhaj et al., 2015). According to [29] (Albu et al., 2021), electrical conductivity is influenced by solution pH, ion valency and degree of ionization. According to [32] (Ghorab et al., 2021), it depends on mineral salts, organic acids and proteins. It can also vary with botanical origin (mono or multifloral nectar). The average values obtained from analysis of the five honeys range from 0.64 ± 0.01 mS∙cm⁻¹ for M3 honey to 1.24 ± 0.012 mS∙cm⁻¹ for M2 honey ( Table 2 ). Only honeys M1 and M4 are not statistically different from each other. Otherwise, all honeys have statistically different electrical conductivities. Honeys M3 and M5 are in line with value of ≤0.8 mS∙cm⁻¹ [2] (Codex Alimentarius, 2022). Honeys M1, M2 and M4 are above the standard. In our study, the variation in conductivity values does not seem to depend on the type of ecosystem.

The results of the present analysis are close to the work carried out by [3] (Koudegnan et al., 2012), whose values for honeys from the plains and mountains of the Guinean zone of Togo vary between 0.26 and 1.39 mS∙cm⁻¹. However, the electrical conductivity values of the five honeys are higher than those of honeys from Algeria, Morocco and Tunisian honeys, which range from 0.12 to 0.41 mS∙cm⁻¹ and Ethiopian honeys which range from 0.4 to 0.43 mS∙cm⁻¹ [6] (Belhaj et al., 2015), [33] (Doukani et al., 2014), [34] (Gebeyehu and Jalata, 2023). Given that the pH values of the five honeys are within the norm, it is difficult to assess the relationship between electrical conductivity and pH in the case of our study.

The density of honey is closely linked to its water content and the quantity and nature of suspended solids (pollen dust, soil dust, fibers, crystallized sugars, wood particles) it contains [35] (Aparna and Rajalakshmi, 1999), [36] (Ayodele et al., 2006). The higher the density of a honey, the less water it contains, and vice versa. Knowing the density of natural honey is a simple and very useful way of detecting poor storage or packaging conditions, or bad commercial practices (counterfeiting). The average density values of the honey samples analyzed ranged from 1.02 (M1, M3 and M5 honeys) to 1.03 (M4 honey) ( Table 2 ). The density of the honeys studied is not statistically different whatever the ecosystem of origin of these Gabonese honeys. On average, the density of the honeys was 1.024. The honey densities found comply with the requirements of the Codex Alimentarius standard for nectar honeys (≤ 1.4). Some authors have also found honey density values lower than or equal to the Codex standard [3] (Koudegnan et al., 2012), [37] (Amir et al., 2010). The more water a honey contains, the less dense it is [38] (Jean-Prost, 1987). However, some authors have published density values for multi-flower honeys that are higher than the Codex standard (1.414 - 1.446, i.e. 1.431 on average) [29] (Albu et al., 2021), [31] (Buveka Ngoma et al., 2018). However, it should be noted that in the case of [29] (Albu et al., 2021) study, the honeys had been analyzed between 3 and 8 years after harvest. In this case, the hypothesis that storage conditions and shelf life had favoured possible evaporation of water from their samples could explain this high-density value for natural multi-floral honey.

The total sugar content of honey is determined by a Brix refractometer in Brix or % equivalent to the quantity of sugars (g) contained in 100 g of honey at 20˚C [19] (Dailly, 2008). Table 2 shows that there is a significant difference (p < 0.05) between the different honeys, with average values ranging from 77% to 82%. The sugar content of honeys M1, M3 and M5 is not statistically different, with an overall average of 78.2%. This is statistically different from honeys M2 and M4, both of which have a total sugar content of 82%. Despite the significant differences observed, total sugar content does not depend on the geographical origin of the honey (savannah-gallery forest complex and primary forest). On the other hand, the low total sugar content can be explained by the high-water content of M5 honey (20.90%). Additionally, the higher total sugar content in honeys M2 and M4 could be due to their low free water content M2 (19.47%) and M4 (17.49%). These total sugar contents are close to the values obtained by [3] (Koudegnan et al., 2012), [10] (Djossou et al., 2013), [21] (Backchiche et al., 2018). These results meet Codex Alimentarius standards, which indicate that honey values exceeding 65% correspond to nectar honeys [2] (Codex Alimentarius, 2022).

The Pearson correlation matrix following confidence intervals of 95% is presented in Table 3 and the principal component analysis (PCA) using the covariance matrix is illustrated in Figure 2 . The correlations between the physico-chemical parameters of the honeys studied shows significant correlations at the 5% threshold. Moisture content is negatively correlated with conductivity (r = −0.628), pH (r = −0.631), density (r = −0.552) and total sugar content (r = −0.890). On the other hand, conductivity is positively correlated with free acidity (0.688) and total sugar content (r = 0.776), and negatively correlated with water content (r = −0.628). These correlations were shown by [3] (Koudegnan et al., 2012) in their work on the analysis of honeys marketed in Togo.

The various correlations obtained in this study are confirmed by principal component analysis (PCA). PCA transforms p inter-correlated quantitative variables into p new uncorrelated variables called principal components [39] (Conti et al., 2007). PCA applied to physico-chemical characteristics yielded two principal components (PC1 and PC2) represented 64.29% of the total variance and the next principal 16.54%, respectively of the concentrated information, i.e. a total of 80.86% of the total variance ( Figure 2 ). PC1 can be linked on the negative side to water content and the positive side to conductivity, total sugar content and pH, while PC2 is linked to density on the positive side.

The distribution of the studied honeys based on the physico-chemical parameters showed two distinct groups. The first group characterized by high free acidity and high electrical conductivity includes M1 (Moulendé 2022), M2 (Libreville 2022) and M4 (Benguia 2019). The second group with high free water content comprises M3 (Moulendé 2019) and M5 (Okondja 2019).

Values in bold are the most significant.