Materials and methods

Study area. This study was carried out in nine meliponaries located in villages and agricultural areas within the main tropical forest types in the three political states of the Yucatan Peninsula in southeastern Mexico (Figure 1). Our sampling strategy encompassed the range of environmental conditions and vegetation types in which meliponaries are found across the peninsula. According to the vegetation classification by ^{Flores-Guido et al. (2010)} and land-cover data for the region by ^{INEGI (2018)}, the meliponaries of the towns of Baca, in Yucatan, and of Pucnachén, Chunkanán, and Lerma, in Campeche, are located in areas of deciduous tropical forests. Meanwhile, the localities of the other two meliponaries of Yucatan, Maní and Dzan, are located in semi-deciduous tropical forests. In addition to native vegetation, there are agricultural areas surrounding the Yucatan meliponaries and pastures near the Campeche meliponaries. On the other hand, the Quintana Roo meliponaries are surrounded by semi-evergreen tropical forest; two of these also include rainfed agriculture (Yaxché and Chanchén Palmar) and the third, cultivated pastures (San Juan de Dios; see Supplementary material Figure S1 for detailed maps of each meliponary).

Figure 1 Map showing the locations of the nine meliponaries studied in the Yucatan Peninsula, Mexico, three in each sociopolitical state (Campeche, Yucatan, and Quintana Roo). 

In general, meliponaries are structures with wooden columns and a palm roof (Figure 2A). There can be low protective walls or fences between the columns that do not limit the flight of bees. The floor can be soil or cement, and water channels are common at the edges to prevent predators from entering the meliponary. Although the traditional method of keeping bee colonies in jobones (hollow tree trunks; ^{Villanueva-Gutiérrez et al. 2005}) is still common, many beekeepers currently use divisible wooden boxes that facilitate managing the colony. In the present study, jobones were found only in the meliponaries of Maní and Chunkanán.

Figure 2 (A) A typical traditional meliponary in the Yucatan Peninsula, Mexico. This shack-like structure is a small bee house that serves as a home to the Mayan bee Melipona beecheii colonies. Most colonies are contained in rectangular wooden hives, while only a few are found in hollow logs called “jobón” (bottom right of the photograph). (B-C) Sampling scheme and picture illustrating the Point-Centred Quarter Method: Sample points (filled circles) were positioned every 10 meters along a 150-meter-long transect. Four quadrants (1, 2, 3, and 4) were outlined at each sampled point. Then, the distance (dashed lines), identity, height, and crown diameters of the nearest plant in each quadrant (open circles) to the sample point were recorded. 

Vegetation characterization. Fieldwork was carried out during the dry season (February to April) of 2022 and 2023. This season was selected because it is the time of the year when most melliferous plant species are blooming and, therefore, when the largest amount of nectar and pollen is collected and accumulated in M. beecheii colonies (^{Di Trani & Villanueva-Gutiérrez 2018}), leading to peak honey production in the Yucatan Peninsula (^{Echazarreta et al. 1997}). This made it possible to detect the species of melliferous plants that blossom in the study zones (^{CONABIO 2016}).

To characterize the vegetation surrounding each meliponary, we traced four 150 meter-long transects running in the direction of the four cardinal points (north, south, east, and west), with the meliponary at the center. In cases where ground conditions or private land ownership did not allow tracing a straight line toward the respective cardinal point, we ensured that each transect was oriented to maintain the longest possible distance from the others, avoiding overlap and ensuring the greatest possible representativeness of vegetation cover in the area.

Fifteen evaluation points were selected along each transect (one every 10 m), and the point-centered quarter method (Figure 2B-C) was followed at each point (^{Brower et al. 1997}, ^{Mitchell 2010}). Compared to other area-independent methods, this method allows for accurate estimations of plant density (^{Morisita 1978}) and increases the sample size with a similar sampling effort (^{Cottam & Curtis 1956}). The point-centered quarter method consists of tracing an imaginary line perpendicular to the transect at each evaluation point, thus creating four quadrants with the central point as the origin; in each quadrant, the plant closest to this central point was sampled (Figure 2B). For each plant, including herbs and climbers, we recorded its taxonomic identity, distance from the quadrant center, two perpendicular diameters of its crown to calculate its crown cover, its height, and whether it was blooming or not. Additionally, for trees and shrubs, we measured the diameter of the trunk in individuals with a diameter at breast height (DBH) of at least 1 cm. The DBH was recorded 1.3 m above the ground, and in trees with multiple stems, DBH was recorded for each stem with DBH of at least 1 cm. The height of each individual was measured in meters with a flexometer up to 2.5 m high; for larger plants, the height was visually estimated, always by the same trained evaluator to ensure consistency of the estimates. Since evaluations of vegetation around the meliponaries may have been carried out at different stages of the blooming period of the species, a given plant was considered to be in the blooming stage if it had at least one flower in any of the two census years. This allowed us to identify the species that were blooming in the days near (before or after) the sampling time.

Taxonomic identification was carried out in the field; when a given specimen could not be identified in situ, samples of branches and flowers or fruits were collected for comparison with voucher specimens deposited in the Herbarium of the Centro de Investigación Científica de Yucatán, A.C. for taxonomic identification.

Analysis of (melliferous) plant species composition. The species composition was analyzed considering all the taxonomic groups in the vegetation surrounding each meliponary, as well as the percentages of blooming species, in order to evaluate the availability of food sources for bees at each study site. We calculated the Importance Value Index (IVI; ^{Curtis & McIntosh 1950}) to analyze the composition of melliferous and non-melliferous species, based on the relative frequency, density, and cover of each species. The index is calculated as the sum of these three metrics in percentage per species, so those with the highest percentages are considered the most important and dominant species. As we used the point-centered quarter method for vegetation assessment, the IVI was calculated with the approximation of ^{Mitchell (2010)}, adapted for this sampling method. The relative frequency was estimated as the proportion of sampling points in which a given species was found; the relative density was estimated as the proportion of quadrants in which a given species was found; and the relative cover was calculated as the proportion of the area occupied by a given species, including herbaceous, climbers, shrubs and trees. We calculated the relative cover using the formula π·r², where r is the radius calculated from the average of the two crown diameters measured divided by 2.

To assess the representativeness of the melliferous species relative to the total number of species recorded per meliponary, we reviewed the lists of melliferous plants, that is, those serving as sources of nectar and pollen, reported for beekeeping and meliponiculture in the Yucatán Peninsula (^{Alfaro-Bates et al. 2010}, ^{Flores-Guido & Vermont-Ricalde 2011}, ^{Quezada-Euán 2018}, ^{Ramírez-Arriaga et al. 2018}, ^{Villanueva-Gutiérrez et al. 2018}). From these studies, 114 melliferous species were identified (Supplementary material Table S1). In addition, to assess the availability of food resources for bees in the study period, blooming and non-blooming plant species were also considered.

Analysis of species richness and diversity indices. The alpha diversity or local diversity of the vegetation around the meliponaries was analyzed using the Hill numbers (^{Hill 1973}). These are a family of diversity indices mathematically unified in a general equation, which differ from each other only by the value of the parameter q:

Where D is the calculated diversity, S is the number of species in the community, and p_i is the relative abundance of species i. The parameter q regulates the sensitivity to the relative abundances of the species (^{Chao et al. 2014}, ^{Jost 2019}). When q = 0, the Hill number of order 0 is calculated, which estimates species richness without considering the abundance of each species. When q = 1, the Hill number of order 1 is estimated, which is the exponential of Shannon’s index and weighs each species exactly by its relative abundance; therefore, it estimates the diversity of equally common species. When q = 2, the Hill number of order 2 is estimated, which is the inverse of Simpson’s index that gives more weight to the most abundant species and, therefore, estimates the number of dominant species. Hill numbers (q = 0, 1, and 2) were calculated using the approximation of ^{Chao et al. (2014)} implemented through the iNEXT package in R (^{Hsieh et al. 2016}, ^{R Core Team 2023}), from which rarefaction/extrapolation curves of species diversity were produced.

To compare the diversity of plant species between meliponary localities by forest type, the sampling coverage was first estimated from the sample size. Sampling coverage indicates the proportion of the species richness of the analyzed community that was recorded through the sampling effort considering the abundance of each species (^{Chao & Jost 2012}). Subsequently, rarefaction/extrapolation curves were constructed based on samples to provide asymptotic diversity estimators based on Hill numbers with their respective 95 % confidence intervals, using a bootstrap method with 200 repetitions. In both cases, that is, sampling coverage and diversity curves, extrapolations were made up to twice the sample size reached (n = 240, extrapolations up to 480 plants), as recommended by ^{Chao et al. (2014)}.

Analysis of the vertical structure of vegetation. Five vegetation strata were established based on the main growth forms recorded (herbaceous, shrub, tree, and climbing) and plant height. The Low vegetation stratum grouped herbaceous species, as well as young trees and climbers less than 1 meter in height; the Middle stratum included shrub species, as well as young trees and climbers between 1 and 2 m in height; and the High stratum comprised only trees and climbing plants that exceeded 2 m in height. Furthermore, for better resolution in the analysis and because the semi-evergreen forest of the peninsula has three vertical strata (^{Flores-Guido & Vermont-Ricalde 2011}), the High stratum was further divided into three height categories: High1: 2 - 10 m, High2: 10 - 15 m, and High3: >15 m.

The vertical structure of the vegetation was analyzed and compared between meliponaries by constructing graphs of the percentage of individuals by stratum. In addition, we analyzed the heterogeneity of the vertical structure of vegetation and the number of dominant strata using the Hill number of order two (q = 2), which gives more weight to the strata with higher frequency (^{Ehbrecht et al. 2016}). Therefore, we used the five strata generated (Low, Middle, High1, High2, and High3) and the number of individuals per stratum to derive the effective number of dominant strata (ENDS) per meliponary. Finally, the mean height in the High stratum was compared between vegetation types using the Kruskal-Wallis non-parametric test, as these data were not normally distributed (Shapiro-Wilk normality test, W = 0.9141, P < 0.001).

Analysis of the horizontal structure of vegetation. Horizontal structure refers to the spatial distribution of individuals and their cover in the plant community. The horizontal structure was estimated using tree density (number of trees per unit area: 1 hectare) and basal area (area of the trunks obtained from DBH, expressed in m² per hectare). In addition, the mean DBH between plant forms and meliponaries was compared using the Kruskal-Wallis non-parametric test since these data were not normally distributed (Shapiro-Wilk normality test, W = 0.81564, P < 0.001). A Dunn test was also used to analyze the differences between meliponaries.

Results

Species composition and diversity. A total of 2,153 individuals were recorded in the 36 transects studied, of which 2,008 individuals were identified at the species level, 119 at the genus level, and 26 at the family level. This resulted in 312 taxa, 250 genera, and 73 families, considering all the localities studied. Meliponaries surrounded by fragments of the original type of vegetation (especially semievergreen tropical forest), rather than agricultural areas or human settlements, tend to have a higher taxonomic richness. For example, Baca and Dzan meliponaries, located in tropical deciduous and semi-deciduous forests, are mainly surrounded by agricultural zones, and showed the lowest taxonomic richness, with 15 families and 22 species, and 21 families and 36 species, respectively. In contrast, the meliponaries of San Juan de Dios and Chanchén Palmar in Quintana Roo, surrounded by relatively well-conserved semi-evergreen tropical forests, exhibited the highest taxonomic richness. A total of 47 families were recorded in each location, with 98 plant species in San Juan de Dios and 79 in Chanchén Palmar (Table 1, Supplementary material Figure S1).

Regarding the representation of blooming species, meliponaries located in semi-deciduous forests attained the highest percentages, with 63.9 and 50.7 % in Dzan and Maní, respectively (Table 1). The lowest percentages were observed in Chunkanán, located in a deciduous forest area (19.5 %), and Yaxché, in a semi-evergreen forest (21.1 %). We found that 50 out of the 114 melliferous plant species reported in the literature were present in the meliponaries. In terms of the blooming melliferous species, none were observed in Yaxché and in Chunkanán, only 1.3 % of plants were blooming melliferous species (a single species: Ipomoea triloba, Convolvulaceae). In all other meliponaries, the percentage of blooming melliferous species varied between 5.1 % in San Juan de Dios and 13.8 % in Dzan. The most common blooming melliferous species were Waltheria indica (Malvaceae) and Tridax procumbens (Asteraceae), present in five and four localities respectively. Croton humilis (Euphorbiaceae), Jacquemontia pentanthos (Convolvulaceae), Piscidia piscipula (Fabaceae), Senna racemosa (Fabaceae), and Viguiera dentata (Asteraceae), were all present in two localities. The remaining 13 blooming melliferous species were recorded in a single locality (Table 1).

Representation of melliferous species. The analysis of vegetation, as a whole, showed that non-melliferous species had a higher IVI than melliferous plants in most meliponary localities. In fact, in eight localities, non-melliferous species accounted for more than 58 % of the total IVI of vegetation, while melliferous species represented 28 to 42 % (Figure 3). The only exception was Dzan, where melliferous species accounted for 62.5 % of the IVI. However, this value was mainly due to the contribution of Citrus x sinensis (19.72 %), a cultivated melliferous species. On the other hand, when considering the blooming vegetation subset, the IVI of the melliferous species was remarkably low. In fact, in the meliponary localities of Yaxché and Chunkanán, the IVI of the blooming melliferous species was 0 and 0.75 %, respectively (Figure 3). In the remaining localities, the IVI did not exceed 15 %, except in the meliponary locality of Dzan (32.4 %), where C. x sinensis was again the main blooming melliferous species. The other blooming melliferous species that stood out were P. piscipula (6.5 %), T. procumbens (5.82 %), and S. racemosa (4.8 %) in the localities of Baca, Maní, and San Juan de Dios, respectively (the list of all recorded plant species and their corresponding IVI is shown in the >Supplemetary material Table S1).

Figure 3 Representation of blooming melliferous and non-melliferous species in the Importance Value Index (IVI) of the plant community surrounding nine meliponaries in the three socio-political states (Yuc = Yucatán, Camp = Campeche, and Q. Roo = Quintana Roo) and in the three main types of vegetation of the Yucatan Peninsula, México. Melliferous and non-melliferous plant species were categorized according to ^{Alfaro-Bates et al. (2010)}, ^{Flores-Guido & Vermont-Ricalde (2011)}, ^{Quezada-Euán (2018)}, ^{Ramírez-Arriaga et al. (2018)} and ^{Villanueva-Gutiérrez et al. (2018)}. 

Species richness and diversity indices. The overall coverage obtained in the sampling was 85.5 %. The lowest sampling coverage was recorded in San Juan de Dios, with 74.6 %, and the highest was observed in Baca, with 95.0 % (Figure 4A). Extrapolations from the sampling coverage curves indicate that doubling the current sampling effort, i.e., using 300 m transects (n = 480), would result in an estimated sampling coverage of 93.28 %.

Figure 4 Diversity parameters based on the Hill numbers and species accumulation curves for plant communities of vegetation surrounding nine meliponaries on the Yucatan Peninsula. Data are presented by socio-political state (Yuc = Yucatán, Camp = Campeche, and Q. Roo = Quintana Roo) and by type of vegetation. (A) Sampling coverage curves for each meliponary area. (B) Comparison of diversity parameters among meliponaries: species richness (q = 0); equally common species diversity (exponential of Shannon's index, q = 1); dominant species diversity (inverse of Simpson's index, q = 2). Diversity curves were constructed using rarefaction (solid lines) and extrapolation (dashed lines) with estimations based on sample size. Shaded areas represent 95% confidence intervals obtained using a Bootstrap method with 200 replications. 

The mean plant species richness (q = 0) in the nine meliponaries evaluated was 70.8 species. The locality with the highest species richness was San Juan de Dios, with 109 effective species, located in a semi-evergreen forest area. In the same vegetation type, the Chanchén Palmar and Yaxché meliponaries showed 89 and 82 species, respectively. However, the species richness in the latter two localities showed no significant differences compared to the Maní meliponary (75 species), located in a semi-deciduous forest area, and compared to the Chunkanán (83 species) and Lerma (72 species) meliponaries, located in deciduous forests (Figure. 4B). In contrast, Baca, also located in a deciduous forest area, presented the lowest species richness with only 25 species.

Concerning the diversity of equally common species (q = 1), the San Juan de Dios meliponary (semi-evergreen forest) was also the most diverse locality with 79 effective species. In contrast, Baca (deciduous forest) and Dzan (semi-deciduous forest) were the least diverse localities with 8 and 16 effective species, respectively. The other localities showed intermediate diversity values that varied between 35 and 58 effective species, and the overlap of their confidence intervals showed no significant differences among vegetation types (Figure. 4B). Regarding the number of dominant species (q = 2), San Juan de Dios again was the locality with the highest value with 58 species. At the opposite end are the Baca and Dzan localities, with 5 and 8 dominant species, respectively (Figure 4B). In the other localities, the overlap of confidence intervals showed similarities in the number of dominant species ranging from 23 to 35, regardless of the vegetation type.

Vertical structure of vegetation. In total, 116 taxa of herbaceous plants, 55 shrubs, 103 trees, 29 climbers, 6 palms, and 3 rosette species were identified. The Low stratum showed the highest percentage of individuals in six of the nine meliponary localities studied, with values exceeding 50 % in Baca, Pucnachén, Chunkanán, Lerma, Maní, and Yaxché (Figure 5). In the other three meliponaries, two of which are found in semi-evergreen forests (Chanchén Palmar and San Juan de Dios), a more even distribution of the percentages was observed between the Low, Middle, and High1 strata. The Dzan meliponary, located in a semi-deciduous forest area, was the only site where the High stratum exceeded 40 % of the recorded individuals. It is worth mentioning that Yaxché was the only locality that presented a plant in the High3 stratum, that is, measuring more than 15 m high (Enterolobium cyclocarpum, 17 m).

Figure 5 Percentage of plants in each vertical stratum in the vegetation surrounding nine meliponaries in the Yucatan Peninsula. Data are presented by socio-political state (Yuc = Yucatán, Camp = Campeche, and Q. Roo = Quintana Roo) and by type of vegetation. The High stratum category was subdivided into High1 - H1 (tress 2-10 m high), High2 - H2 (trees 10-15 m high), and High3 - H3 (trees > 15 m high). The effective number of dominant strata (ENDS), based on the abundance per species in each stratum (q = 2, Hill number), is shown for each meliponary. 

Although the percentage values indicate that the Low stratum prevails in most meliponary localities, the effective number of dominant strata (ENDS), derived from the Hill number of order 2, shows a different perspective (Figure 5). The analysis of these data highlights that only two localities, Baca (deciduous forest) and Yaxché (semi-evergreen forest), recorded less than two dominant strata, while the ENDS ranged from 2.1 to 2.8 in six meliponaries. Only San Juan de Dios (semi-evergreen forest) showed three dominant strata, with a more balanced distribution between the Low, Middle, and High strata, as mentioned above.

The analysis of species composition and abundance by stratum showed that, in general terms, the most abundant species in the Low stratum was Parthenium hysterophorus (Asteraceae), a non-melliferous plant, with a total of 90 plants. No melliferous species reached this abundance in this stratum (Table 2). In fact, the most abundant melliferous species in the Low stratum never exceeded 30 individuals. Among them, T. procumbens had 28 plants in Maní and W. indica had 21 in Pucnachén. In addition, these two were the most common melliferous species, recorded in two meliponaries each. At the opposite end, in the meliponary locality of Baca, no melliferous species were found in the Low stratum (Table 2). Interestingly, three tree species were identified among the most abundant melliferous species in the Low stratum: Cedrela odorata and Metopium brownei in San Juan de Dios and P. piscipula in Dzan.

As for the Middle stratum, the most abundant species was Hamelia patens, with 53 individuals. However, no melliferous species reached such abundance in the Middle stratum, the most abundant being the cultivated species C x sinensis, with 32 individuals, all in the vegetation of the Dzan meliponary (Table 2). In the rest of the meliponary localities, melliferous species in the Middle stratum were not very abundant, with numbers ranging between one and seven plants.

On the other hand, the High stratum had melliferous individuals in the three levels. However, the tree stratum in meliponary localities generally exhibited modest dimensions, with a mean height of 5.5 ± SD 2.9 m in the High stratum. The vegetation type with the highest trees was the semi-evergreen forest (6.3 ± SD 3.0 m), which differed from the deciduous forest (5.2 ± SD 2.6 m) and the semi-deciduous forest (5.2 ± SD 2.9 m), but the latter two did not differ from each other (Kruskal-Wallis rank sum test: χ² = 15.23, df = 2, P < 0.001). Although 468 tree individuals were detected in the High stratum, only 40 individuals larger than 10 m in height were recorded, representing a mere 8.6 %. Of the total, 261 melliferous plants belonging to 32 species were identified, but only 22 plants (11 species) reached heights above 10 m. Neomillspaughia emarginata was one of the most abundant melliferous tree species, with 45 individuals in Dzan (Table 2). For the other species, the number of plants did not exceed 10 individuals, except for Senna atomaria, with 11 individuals in Baca.

Horizontal structure of vegetation. The tree vegetation of the meliponaries was characterized by a global mean DBH of 7.04 ± SD 3.2 cm. The comparison of vegetation types showed no differences in the DBH of trees between the deciduous forest (7.28 ± SD 6.6 cm) and the semi-evergreen forest (9.2 ± SD 5.8 cm), but the latter showed higher values than the semi-deciduous tropical forest (5.7 ± SD 2.9 cm; Kruskal-Wallis rank sum test: χ² = 24.21, df = 2, P < 0.001). This variation was also reflected at the level of individual meliponaries (Kruskal-Wallis rank sum test: χ² = 54.25, df = 2, P < 0.001); the greatest difference was observed between Dzan and Yaxché, with mean DBH values of 5.34 (± SD 0.51) cm and 11.78 (± SD 2.15) cm, respectively (Table 3).

Regarding tree density in meliponaries, a global average of 4,496 (± 4,089) trees per hectare was recorded, with no significant differences among vegetation types (Kruskal-Wallis rank sum test: χ² = 8.35, df = 2, P = 0.015). The Lerma and Baca meliponaries, both in deciduous forests, exhibited the highest densities with 9931 (SD ± 4,344) and 8,724 (SD ± 3,400) trees/ha, respectively. In contrast, San Juan de Dios, located in a semi-evergreen forest, showed the lowest density with 812 (SD ± 1059) trees/ha. We observed extreme values of tree basal area between meliponaries. Specifically, Baca recorded the highest value with 53.1 m²/ha, followed by Lerma with 30.8 m²/ha, while San Juan de Dios had the lowest value with 2.7 m²/ha (Table 3). The global mean for the basal area of trees in the vegetation surrounding meliponaries was 23.4 (± SD 21.2) m²/ha.

Discussion

Our results suggest that the impact of human activities, had a more pronounced effect on vegetation structure and diversity compared to the surrounding forest type, contrary to expectations. Since the meliponaries are located near human settlements, activities such as agriculture, horticulture, and urbanization may had an important effect on plant and structure diversity. The influence of human activities was evident in the mean tree height of trees in the high stratum of vegetation around the meliponaries, which was 5.5 m much lower than the reported tree heights in well-conserved deciduous (8-12 m), semi-deciduous (13-18 m), and semi-evergreen tropical (15-25 m) forests in the Yucatan Peninsula (^{Durán & Olmsted 1999}, ^{Flores-Guido et al. 2010}). An in-depth analysis of the shifts in land use correlated with the surrounding vegetation of meliponaries will significantly bolster our findings.

We also found some evidence suggestive of an effect of biotic homogenization. This phenomenon occurs when human impacts (e.g., deforestation) in the ecosystem promote the replacement of distinctive local species with others more resilient to disturbances or when the ecosystem conditions become more favorable for widely distributed species (^{McKinney 2006}). We observed ruderal species, such as P. hysterophorus or T. procumbens, as well as widely distributed species, such as P. piscipula and C. odorata (both of economic interest), in the vegetation surrounding almost all meliponaries. Furthermore, we found crops, such as citrus fruits, in several of the meliponary localities, indicating a targeted selection of plants in areas adjacent to the meliponaries. The similarity in the effective number of equally common species (Hill number of order 1, q = 1) observed in six of the nine meliponaries studied encompassing different vegetation types further suggests an effect of biotic homogenization.

Despite the similarities in the diversity between meliponaries, we observed the expected pattern of higher plant diversity in meliponaries located in semi-evergreen forests compared to other forest types, but only when comparing Baca and San Juan de Dios. These localities are in deciduous and semi-evergreen forests, respectively, characterized by contrasting precipitation levels (lower in Baca and higher in San Juan de Dios), which is consistent with the expected dissimilarities in species diversity between different types of vegetation (^{Flores & Espejel 1994}, ^{Flores-Guido et al. 2010}). On the other hand, the low species diversity observed in the Dzan meliponary, located in a semi-deciduous forest (with intermediate precipitation levels) can be attributed to human activities rather than forest type since this meliponary was surrounded by a plantation of C. x sinensis.

The high vegetation stratum around the meliponaries showed a mean height of 5.5 m, a mean DBH of 7.04 cm, and a mean density of 4,496 (± 4,089) trees/ha. These values are characteristic of young secondary vegetation, which has high densities of small trees (^{Denslow & Guzman 2000}, ^{Guariguata & Ostertag 2001}). Similarly, in Puerto Rico, ^{Aide et al. (1996)} reported tree heights of up to 5 m and a density of 4,000 trees/ha in secondary vegetation with 9.5 to 15 years of agricultural land abandonment. Additionally, in a semi-deciduous tropical forest of the Yucatan Peninsula, ^{Saenz-Pedroza et al. (2020)} found that the density of trees with DBH of 1 cm to 5 cm decreased from 12,000 to 6,000 trees/ha in areas with between 5 and 60 years of abandonment, while the density of trees with DBH > 5 cm increased from > 500 to 1,500 trees/ha in the same successional age range. Furthermore, considering compositional characteristics, we observed the presence of Gymnopodium floribundum, M. brownei, and Bursera simaruba mainly in the meliponaries of Quintana Roo. These three species, along with Sabal yapa (a genus recorded in our transects) and Hampea trilobata (undetected in the present study), form a plant community typical of early abandoned secondary vegetation in Quintana Roo State. Specifically, this community is associated with young secondary vegetation, 8 to 10 years of development (^{Sánchez & Islebe 2000}).

The diversity of plants, including both melliferous and non-melliferous species, and their abundance and distribution by strata, suggest that the vegetation surrounding meliponaries is constantly influenced by human activities and may not ensure an adequate food supply for M. beecheii colonies. However, a study conducted by ^{González-Avilés et al. (2023)} in the Sierra del Abra Tanchipa Biosphere Reserve, San Luis Potosí, Mexico, found that disturbed areas of vegetation have a high richness of melliferous plant species. These areas consist of a mixture of both native and introduced melliferous species, which can serve as a rich source of food for bees, particularly Apis mellifera.

Although there are few studies on plant species used by M. beecheii as food sources in the Yucatan Peninsula, it has been observed that this bee species can visit approximately 114 species of melliferous plants in the region (^{Alfaro-Bates et al. 2010}, ^{Flores-Guido & Vermont-Ricalde 2011}, ^{Quezada-Euán 2018}, ^{Ramírez-Arriaga et al. 2018}, ^{Villanueva-Gutiérrez et al. 2018}). These 114 plant species accounted for less than 43 % of the IVI of the surrounding vegetation in eight of the nine meliponaries evaluated. The exception was the Dzan meliponary, where the IVI of melliferous plants reached 62.5 %. However, this high value is largely due to the presence of C. x sinensis; if this species is excluded, the IVI of melliferous plants around the Dzan meliponary would be similar to values observed in other sites (42.78 %).

In addition to the low IVI values of melliferous plants around the meliponaries, the structure and young successional age of vegetation may limit food availability for bees, since some species of trees can take decades to start blooming (^{Hackett 1987}). In addition, the age at first blooming is generally related to the final plant size; thus, the larger the tree, the larger the size it should reach to undergo its first blooming (^{Thomas 1996}). Therefore, if the vegetation is young or the trees are cut and not allowed to grow, as in disturbed vegetation, the number of individuals that manage to flower is reduced, thus limiting food availability for bees. The low representation of blooming melliferous plants during the sampling time, which was carried out in the season of highest honey production in the Yucatan Peninsula agrees with this interpretation. Indeed, no blooming melliferous plants were observed in one meliponary and another five meliponaries had IVI values lower than 9 %.

The availability (abundance and species richness) of plants that produce nectar or pollen is essential to ensure sufficient food not only for M. beecheii but also for the other species of bees with which it coexists and competes. Additionally, the studies carried out to date evaluating the food resources used by M. beecheii have analyzed pollen collected by these bees but have not considered the surrounding vegetation; the response of this bee species to variations in the supply of food resources in the local ecosystem has not been addressed. Assessing the relationship between the diversity of available food sources in these meliponaries and those actually used by M. beecheii colonies is highly warranted (see ^{González-Avilés et al. 2023}).

There are two caveats to this study: the possible misclassification of some plant species as non-melliferous and the lack of information on the flight radius of bees. Some of the species may have been misclassified as non-melliferous in this work based on the literature on the melliferous flora of the Yucatan Peninsula, which may lead to underestimating the percentage of melliferous plants within the sampled flora. However, our list includes species such as G. floribundum, P. piscipula, V. dentata, Lysiloma latisiliquum, and B. simaruba, which are among the main melliferous species in the region in terms of the volume of food supplied to bees (^{Alfaro-Bates et al. 2010}, ^{Flores-Guido & Vermont-Ricalde 2011}) and are widely represented in the forests of the Yucatan Peninsula (^{Miranda & Hernández-X. 1963}, ²⁰¹⁴, ^{Rzedowski & Huerta 1994}). Furthermore, our list covers species with different blooming periods, so it is possible that some of the species misclassified as non-melliferous are actually poorly represented melliferous species of little melliferous importance.

On the other hand, no data are currently available on the flight radius of M. beecheii. For other species of the same genus, the flight radius varies as follows: 1,120 m in Melipona subnitida (^{Silva et al. 2014}), 1,800 m in Melipona mandacaia (^{Kuhn-Neto et al. 2009}), and 2,470 m in Melipona fasciculata (^{Kerr 1987}). This raises the possibility that the floristic diversity available to the bees studied here may have been underestimated, since our sampling was carried out within a radius of 150 m around the meliponaries. However, we believe that our results are indicative of the prevailing conditions in the peninsula due to its physiographic homogeneity, the widespread presence of the main agricultural activities (^{Flores & Espejel 1994}, ^{Duno-de Stefano et al. 2018}), and the predominance of secondary vegetation (^{Islebe et al. 2015}). These characteristics support the extrapolation of our results to greater distances around the meliponaries studied and other meliponaries of the Yucatan Peninsula.

To summarize, our results suggest that vegetation type has a low impact on the composition and structure of vegetation surrounding meliponaries. Rather, the influence of human activities, which was not directly assessed in this study, appears to be a more important factor. The low height of trees and high abundance of individuals in the low stratum indicate that the vegetation surrounding meliponaries is in early successional stages, possibly due to human impacts derived from agriculture and urbanization. However, it is necessary to analyze the relationship between current land use and the diversity and availability of plant species that serve as bee food resources. The availability of melliferous plant species observed around meliponaries cannot be fully explained by the analysis of vegetation structure and species diversity and composition alone. The effects of changes in land use on food availability for M. beecheii should be investigated to provide a more comprehensive explanation, as well as well-supported guidelines to foster meliponiculture in the Yucatan Peninsula.