2.1 Propolis

The study focused on a collection of propolis collected throughout Uruguay during the 2020-2021 season. The location of the apiaries is depicted in Figure 1. There are 14 locations with harvests in different seasons, totaling 22 propolis samples, each with 5 repetitions, as detailed in Table 1. An attempt was made to cover the diversity of soil units and environments present in the country. Thus, apiaries were located on basalt, sandstones, quaternary sediments, crystalline basement, hill ranges and coastal soils. Propolis was manually collected from beehive frames after 5 minutes of freezing at -20 °C, removing residues from wax and bees, and then preserved at room temperature²⁶.

2.2 Characterization of the plant environment in each locality

Using information from the Ministry of Livestock, Agriculture and Fisheries²⁹, a radius of 2 km was established around each apiary and the surface area of each of the land use categories was measured. The locality of Punta Colorada was initially assigned 100% to the Livestock category with over 10% artificial pasture. However, Figure 2 illustrates the actual situation of this apiary. To correct this, the sea and coastal area were excluded from the total. The new 100% is distributed between 54% livestock with over 10% artificial pasture and 46% corresponding to the urban environment, which is added to Table 2. The results were expressed as a percentage of the total area, 1258 ha. The different categories and percentages are detailed in Table 2.

Figure 3 shows an example of the three layers of information (satellite image with apiary location, 2 km flight radius, and land use areas).

2.3 Polyphenol content

From each propolis sample, 2 g were taken and diluted in 20 ml of 80% ethanol. It was then stirred in darkness for 4 h at 30 °C, using a thermoagitator (Gyrotory® New Brunswick Scientific Co. Edison, N, J, USA). It was filtered using ashless paper filters MN 640 d N°42 (Macherey Nagel), and stored in a refrigerator (4 °C).

The following standards were used for the determination of polyphenols: apigenin (4',5,7-trihydroxyflavone), umbelliferone (≥ 98%, Sigma-Aldrich, USA), caffeic acid (≥ 98%, Sigma-Aldrich, USA), p-coumaric acid (≥ 98%, Sigma-Aldrich, USA), CAPE (≥ 97%, Merck Millipore Corp., USA), trans-ferulic acid (99%, Sigma-Aldrich, USA), boldine (natural from Peumus boldus molina, ≥ 98%, Supelco, USA), chrysin (≥ 96.5%, Sigma-Aldrich, USA), galangin (≥ 95% Sigma-Aldrich, USA), kaempferol (≥ 97%, Sigma-Aldrich, USA), morin (phyproof ® Reference Substance, ≥ 95%, Phytolab, Germany), pinocembrin (≥ 95%, Sigma-Aldrich, USA), rutin hydrate (≥ 94%, Sigma-Aldrich, USA), quercetin (≥ 98%, Sigma-Aldrich, USA) and vanillin (Reagent Plus®, 99%, Sigma-Aldrich, USA).

Propolis extracts were analyzed using high-performance liquid chromatography. The equipment used was an HPLC Prominence LC-20A (Shimadzu Corporation, Japan) equipped with autosampler, quaternary pump and diode array detector (SPD-M20A). The separation of each hydroalcoholic extract of propolis was carried out in a Luna C18 column (250 × 4.6 mm, 5 μm, Phenomenex, USA), thermostatized at 35 ºC. The total running time was 60 minutes according to the elution gradient, which consisted of two mobile phases: 1) Water with 0.1% trifluoroacetic (TFA) (Mobile Phase A), and 2) Acetonitrile with 0.1% TFA (Mobile Phase B). The following mobile phase gradients were used at a flow rate of 0.7 mL/min: 0 min 35% B, 10-15 min 50% B, 40 min 80% B, 50-60 min 35% of B. The injection volume was 10 μL. It was quantified according to the external standard method, generating calibration curves for each of the standards. MilliQ-quality water, HPLC-quality acetonitrile (Merck, Germany) and HPLC-quality TFA (≥ 99%, Fisher Chemical, USA) were used. The limit of detection (LD) was 0.04 mg/kg and the limit of quantification (LQ) was 0.12 mg/kg. Results are expressed as grams of polyphenols per kilogram of propolis.

2.4 Minerals

Propolis samples (2 g) were placed in porcelain crucibles with muffle lids with a temperature ramp and a maximum of 580 ºC until white ash was obtained³⁰. The obtained ashes were dissolved in acid solution, according to Tosic and others¹², with modifications. Ultrapure HNO₃ (1M, 65% Merck, sub-boiling distillate) and HCl (6M, Merck, ACS) were used in an Erlenmeyer flask with a steam trap device, on a hot plate (Thermolyne) until complete dissolution. The solution was then filtered using ashless paper filters MN 640 d N°42 (Macherey Nagel) and brought to a volume of 25 ml with MilliQ ultrapure water in a volumetric flask³¹. The content of microelements: Copper (Cu), Iron (Fe), Manganese (Mn) and Zinc (Zn) in propolis was quantified with an Atomic Absorption Spectrometer (Perkin Elmer, AAnalyst 300, USA) equipped with a deuterium lamp as background corrector, with flame (air-acetylene; 8.0 l/min and 1.4 l/min) for Fe, Zn, Cu, Mn. A standard curve was prepared for each analyte from single-element solutions of Cu, Fe, Mn and Zn, containing 1000 mg/l (Perkin Elmer or Fluka). Samples and standards were diluted with deionized H₂0 and contained 0.5% ultrapure HNO₃. The content of macroelements calcium (Ca), magnesium (Mg), sodium (Na) and potassium (K) was quantified in the same ash solution described above. A Lumina multielement lamp (Perkin Elmer, USA) was used for Ca (422.7 wavelengths) and Mg (285.2 wavelengths), while Na and K were measured with emission (589 and 404.4 wavelengths, respectively). A standard curve was prepared for each analyte from solutions of Ca, Mg, Na and K of 1000 mg/l (CertipurR, Merck, Germany). A blank (acids only) was run with the samples. Data were expressed in milligrams of each mineral per kilogram of propolis (mg/kg). The limit of quantification (LQ) was calculated as the concentration resulting from 10 times the blank deviation, and the limit of detection (LD) was calculated as the concentration resulting from 3.3 times the blank deviation. The LQ values for Ca, K, Mg, and Na were 5.43, 0.06, 0.75, and 0.11 mg/l, respectively, and the LD values were 1.79, 0.02, 0.25, and 0.04 ml/l, respectively. For Cu, Fe, Mn and Zn the LQ was 0.01, 0.03, 0.03 and 0.01 mg/l, respectively, and the LD was 0.004, 0.01, 0.01 and 0.003 mg/l, respectively. The materials used in the preparation of samples and solutions were conditioned with a combination of the four factors of the Sinner circle: mechanical action (non-metallic brushes), and chemical, with times and temperatures according to the material. Soaking was used for 24 or 48 hours in deionized H₂0 with 10% or 25% concentrated HNO₃, and finally rinsed several times with deionized H₂0, following the main procedures of Ballinger & Shugar³² and EPA³³. The materials were air-dried upside down on a clean basket without a towel or stove to prevent contamination. The materials were periodically renewed to reduce contamination.

2.5 Statistical analysis

To compare propolis, an analysis of variance was performed and the Tukey-Kramer test was applied. Propolis samples were compared for their similarity in both polyphenol and mineral content using multivariate analysis techniques. The localities were also compared in terms of their vegetation according to the land use categories. Dendrograms were constructed by standardizing the data using Ward's method and employing Euclidean distance. Correlations between polyphenols and between minerals were calculated. Polyphenols correlated with Pearson's p-values ≥ 0.7 were used to perform a network analysis using Louvain's algorithm³⁴. Rstudio³⁵ and Infostat³⁶ were used.

3.1 Plant environment

Ten out of the 16 categories provided by the MGAP information were detected²⁹. The percentage values of each of them were used to create the dendrogram in Figure 4. In terms of percentages, the most similar localities are Minas, Treinta y Tres and Valle Lunarejo (100%). Cerro Largo closely resembles this group with 93%. Another similarity is observed between Cerrillos and Florida North (77%), and they are 60% similar to Colonia. The locality of Punta Colorada stands alone due to the percentage of the Urban category in land use.

3.2 Polyphenols

The detected and quantified polyphenols are presented in Table 3, 4, 5 and 6. Except for rutin, all others were detected and quantified in some of the 110 propolis samples. Boldine contents are 1000 times lower than other polyphenols, which is why they are expressed in mg/kg. Umbelliferone only appears in propolis from Paysandú and Cerro Largo and in some samples from Treinta y Tres and Valle Lunarejo. Other polyphenols that were not found in some of the 110 propolis include caffeic acid, ferulic acid, CAPE, morin, quercetin and vanillin. The locality with the lowest diversity was Rocha in spring, where boldine, morin, quercetin, umbelliferone or vanillin were not detected. Cerro Largo was the locality with the lowest total amount, while Juanicó presented the highest amounts, in all seasons and especially in summer. The similarity in propolis based on polyphenol content and profile is shown in Figure 5. Except for the propolis from both seasons of Paysandú (Pp and Ps) which appear in different groups, the other localities with more than one harvest (Cerrillos, Juanicó, Florida West and Rocha) present similar propolis in all harvests. In terms of percentage, propolis from Colonia were similar in 88%, those from Juanicó in 84%, Rocha 83% and Cerrillos 78%, while samples from Paysandú were only 65% similar. Graphs present the average polyphenol content of the southwest (Figure 6) and northeast (Figure 7) localities. Ten times less content is observed in the northeast, but the main polyphenols (chrysin, galangin and pinocembrin) are the same, although they decrease proportionally, while quercetin increases compared to the main polyphenols. Based on the three resulting groups (Figure 5), the correlation between the quantified polyphenols within each group was calculated. Only those values of Pearson's coefficient equal to or over 0.7 were used and are presented in Table 7, 8 and 9. Networks in Figure 8, 9 and 10 were constructed with the values obtained for each group.

3.3 Minerals

All analyzed minerals were quantified in all propolis samples. Results are shown in Table 10. The similarity between propolis based on their mineral content is observed in Figure 11. Heterogeneity of compositions is observed, not only between localities but also within the same locality between stations. For example, the propolis from both Juanicó autumn and Cerrillos autumn appear close to each other, but distant from the propolis samples of Juanicó and Cerrillos in spring and summer. The same happens with Paysandú in spring and summer, which appear to be very different. On the other hand, there are cases of very distant localities with similar profiles (Valle Lunarejo and Cerro Largo). R2 values between minerals were calculated. Only Pearson's values over 0.7 were considered, finding a value of R2=0.81 between Zn and Fe, and R2=0.62 between Zn and Mn.

4.1 Plant environment

The 10 land-use categories²⁹ are more than the 8 generated using satellite photos²⁶. In addition, the Urban category was added to achieve a better description of land use. However, there was no improvement in describing different environments to establish an appropriate relationship with propolis based on their polyphenol content (Figure 4 and Figure 5). In pastoral landuse (where low-polyphenol propolis is produced) it is possible to group environments and propolis by their polyphenol content, as shown in red in Figure 4 and Figure 5. Pastoral use could imply the absence of trees or shrubs that provide resins, which could explain the low polyphenol content in these propolis samples. The similarity between Juanicó and Rocha lies in both having a single category of landuse, even though they are different. This is a deficiency of the method employed as descriptive statistics, which simply takes the number as a descriptor. Although there are more categories, they do not better describe the plant landscape. Some localities correspond entirely to one land-use category (Juanicó, Valle Lunarejo, Cerro Largo) when the diversity in these same localities is greater and is better reflected using categories obtained from satellite images.

4.2 Grouping of localities by region according to polyphenols

The polyphenol content allows grouping propolis into three groups. Group 1 is in red (northeast region) in the dendrogram in Figure 4, group 2 in green (combining the three harvest seasons of Juanicó and Paysandú summer), and group 3 in blue (southwest region). Thus, it can be stated that there are two major regions of propolis in the country according to their polyphenol content and profile: the northeastern region with the lowest polyphenol content and diversity (group 1) and the southwest region with the highest content and diversity (groups 2 and 3). In a central area (localities FW and FN, groups 1 and 3, respectively) intermediate values are observed (Table 3, 4, 5 and 6), which are lower than those in groups 2 and 3 FN and higher than group 1 FW. These locations could be in a transitional zone. These regions are presented in Figure 12. The differences in polyphenol content and profile between these groups and within them are explained by variations in soils, climate, and vegetation, as reported in other studies ^_37)(38) . The identified correlations between polyphenols, in groups 2 and 3, may indicate similar plant sources in the southwest region and different plant sources in the northeast region. In all propolis samples, more than 70% of the polyphenols are due to the combined content of chrysin, galangin and pinocembrin, except in Cerro Largo, Rocha spring and Treinta y Tres, where they still represent more than 50%. Bees seem to have a strong selectivity for resins rich in these compounds. Similar percentage values of pinocembrin are reported in Hungary³⁹⁾ and Europe in general¹⁶, and in the case of Hungary, the same happens with chrysin values. In Turkey, variable percentage values of pinocembrin and galangin are observed, but not all localities have high percentages⁴⁰. In Brazil, on the other hand, the share of pinocembrin and chrysin in propolis is lower, with a notable content of hydroxycinnamic acids⁴¹.

4.3 Seasonal differences

In the localities where propolis collection occurred once a year, it was not possible to measure seasonal variations. The slow deposition in these localities indicates the absence of significant plant sources for bees, making them the localities with lower polyphenol content and a poorer profile (Table 1). In Brazil (São Paulo and Minas Gerais), where the climate allows bees to work year-round, variations in resin collection were observed. There were even instances of total absence of resin collection, depending on the environmental and nutritional conditions of bees⁴².

In localities with 2 or 3 harvests, propolis from spring, summer and autumn have similar polyphenol contents, ranging from 78% to 88%. These variations could be explained by changes in the annual vegetation cycle. A study on propolis from Paraná (Brazil) found similar polyphenol profiles between different seasons and years⁴³, explained by the similarity in the vegetation. Variations observed are in the quantities of polyphenols, attributed to changes in temperature conditions affecting both bees and plant physiology in their ability to generate resins. It is also influenced by the nutritional status of colonies, dependent on climatic conditions, affecting the bees' ability to collect resins⁴³. Similar findings were reported in Sonora, Mexico, where propolis exhibited consistent characteristics regardless of the season, with variations mainly in the quantity collected⁴⁴. Generally, seasonal variations occur in cases of marked changes in temperature or rainfall that alter the flora. In Uruguay, the production of resins appears to have more continuity throughout the season than the production of nectar, which is limited to the flowering period. However, two exceptions were found in the propolis of Paysandú spring (Pp) and summer (Ps). They show a change in the profile and concentration of polyphenols, explaining their placement in different groups in the dendrogram (Figure 5) with a similarity of 65%. The polyphenol content in Ps is higher, especially in the top 3 constituents. Only p-coumaric acid, kaempferol and quercetin exhibit reduced quantities in Ps. One hypothesis would be that two different plant sources, the one that supplies p-coumaric acid and kaempferol, on the one hand, and the one that supplies kaempferol and quercetin, on the other, stop producing in summer or are replaced by other plant sources preferred by bees or a combination of these events. The Paysandú apiary is close to the Uruguay River, which acts as a route for plant species from southern Brazil (subtropical). In addition, this locality has higher temperatures than localities in the south. Exotic species can settle and bees could take advantage of them in summer. The differences between polyphenols are shown in Figure 13 and Figure 14, where the three main polyphenols were removed to facilitate the visualization of the others.

4.4 Comparison with other propolis in the world

When comparing the values found with other global propolis polyphenol content, a significant similarity in content and profile is observed with Chinese propolis originating from Populus sp.⁴⁵. Chinese propolis of this origin has a higher content of hydroxycinnamic acids, but different solvents are used⁴⁵, potentially resulting in different profiles. Hydroxycinnamic acids show higher maximum values in propolis from Hungary³⁹, Turkey⁴⁰⁾ and eastern and southwestern Europe¹⁶. In the case of propolis from group 1 (northeast), the values of this family are even lower. This region borders Brazil, where propolis with very low polyphenol content is reported⁴⁶. The similarity in climate, soil and vegetation with southern Brazil explains the results. In Brazil, brown, green and red propolis are analyzed for their polyphenol content,with rutin detected in all of them, unlike in Uruguay. Green propolis shows higher values of caffeic acid and p-coumaric acid, but other polyphenols exhibit similar (kaempferol) or lower (apigenin, chrysin, pinocembrin, and vanillin) values⁴⁶. On the other hand, the contents of chrysin, galangin and pinocembrin cited for Chinese propolis⁴⁵ and European propolis ^_16)(39) fall within the range with few values higher than those found in southwest Uruguay. Soil fertility and greater colonization of European species, whether ornamental or associated with crops, would explain these contents.

4.5 Correlations between polyphenols

Regarding the correlations found between polyphenols, it can be assumed that those correlated come from the same plant species. In this regard, clear associations are observed, such as umbelliferone and vanillin in group 1, or ferulic acid and vanillin in group 3, which could represent different plants in each region. The species Parkinsonia aculeata is reported as a source of umbelliferone when its methanolic extract is studied⁴⁷. This species is present in Uruguay, specifically along the Uruguay River (Paysandú locality)⁴⁸. Other species cited as sources of umbelliferone⁴⁹⁾ are also present in Uruguay. This study mentions, among other species, Picea abies and Platanus acerifolia, both present in Uruguay as introduced species⁵⁰. It also mentions the species Hidrangea macrophylla, also present in Uruguay⁵¹. A comprehensive survey of the flora of each locality far exceeds the objectives of this study. The mention of these species is not intended to attribute the exclusive origin of the found polyphenols to them, other possible sources should be considered. Associations between apigenin, chrysin, galangin, and morin are also observed in both groups 2 and 3 (Figure 9 and Figure 10), both groups in the same region (southwest), which could be due to the same plant origin. There are other more complex associations where this cannot be asserted, requiring further investigation, such as the association between caffeic acid, CAPE, and pinocembrin, observed in group 2 but not in group 3 (Figure 9 and Figure 10). On the other hand, those polyphenols with low correlations with all others could be assumed to originate from different plant sources, such as boldine. This should be taken as an indication of plant origin and further research into polyphenolic compositions in different plants in each region should be conducted. Regarding pinocembrin, one of the most abundant polyphenols present in all the propolis obtained, plant sources of the genera Populus, Parkinsonia and Myrceugenia are mentioned⁵². All these families or genera are present in Uruguay supporting the idea of finding propolis similar to European ones, but with polyphenols provided by native flora. Chrysin, galangin and pinocembrin are mentioned as responsible for different medicinal properties⁵², especially in different types of cancer due to their involvement in apoptosis mechanisms and prevention of angiogenesis⁵³. Group 2 and 3 propolis exhibit promising polyphenol values in terms of their medicinal potential.

4.6 Minerals

The difference between localities in concentration and mineral profile is explained by the different geological materials that originate in the soils in each locality. Additionally, the mineral content of the water that the bees bring to the hives is variable. Differences between studied regions are reported in India, Spain, and Serbia, with sometimes greater differences observed between closer localities ^{_11)(12) 13}. More difficult to explain is the variation within the same locality between seasons (Figure 11). Propolis from the same locality with the same soils and vegetation should be similar in mineral content. In Turkey, propolis from nearby regions (Ardahan and Erzurum) do not show differences in minerals, attributed to their geographical proximity⁵⁴. The minerals in propolis are explained by those naturally present in resins, those added by bees in the process and others that arrive as external contaminants. The latter are variable and depend on environmental factors.

In this regard, the variation in winds and rainfall patterns could explain the arrival or absence of dust to the resins, which, depending on their physical nature, will either retain or not. Rainfall, based on its quantity, washes away these minerals, depending on their solubility. Finally, the different structures of the plants where resins are generated, whether exposed or not, can alter the retention by wind and washing by rain. In terms of concentrations, the analyzed minerals fall within the ranges reported for other propolis in the world ^{_11)(12) 13}. Calcium (Ca) and Potassium (K) are the predominant minerals in most localities, except Treinta y Tres and Rocha (Table 10). In these two localities, iron, magnesium, manganese and zinc show high values that differ from the rest of the localities but fall within reported ranges ^{_11)(12) 13}. The presence of different minerals is a reflection of the soils and plants present in the environment ^_12)(13) . The Treinta y Tres apiary is located on the batholith of the Cuchilla de Dionisio formation. This formation contains minerals such as quartz, feldspars and biotite, which could be the source of iron that explains the high values⁵⁵. The correlation between Fe, Zn and Mn suggests a common source of origin. The Na value in Punta Colorada, while not different from others mentioned, and not statistically higher than other localities, is explained by the proximity to the sea. This element can be considered as a contaminant that arrives from the sea depending on the rainfall and wind patterns⁵⁶. In some studies, the nutritional value of some elements present in propolis (Fe, Mg, Mn, Zn, among others) is highlighted ^{_11)(12) 13}. Without intending to minimize these contributions, valid in certain dietary situations, in Uruguay these minerals are obtained from other food sources. The results suggest that Fe and Zn could serve as geographical markers for the locality of Treinta y Tres.