Description of the study area

The state of Tabasco is located geographically southeastern Mexico between 17° 15' 00'' and 18° 39' 07'' north latitude, 90° 50' 23'' and 94° 07' 49'' west longitude. Limit to the north with the Gulf of Mexico, northwest with the state of Campeche, in the southeast with the Republic of Guatemala and south and west by the states of Chiapas and Veracruz, respectively. The state of Tabasco has five sub-regions characterized by different geographical, ecological and socioeconomic conditions: Chontalpa Centro, Sierra, Rivers and Swamps. In this research samples were collected in various apiaries located in the five sub-regions suitable for the development of apiculture according to SAGARPA (2004).

Figure 1 Map from the state of Tabasco showing the six municipalities where the 38 honey samples were collected. 

Field work

During 2006, 2007 and 2008 the fieldwork was conducted in six apiaries located in five regions, sampling only six municipalities within them. In each apiary collected sedimented honey samples in plastic bottles and stainless steel, using plastic bottles with a capacity of 200 ml. The honey contained in the bottles was previously centrifuged, settled and filtered so that it is free of impurities, according to the technique from apiarist. Each sample was identified with a label with the name of the municipality, apiarist, date of collection and region of origin (Table 1).

Table 1 Identification of honey samples. 

The samples were stored in the Laboratory of Food Science Campus Tabasco from the Postgraduate College in Agricultural Sciences, until analysis and identification.

Laboratory analysis

Absolute number of pollen grains

To calculate the total amount of pollen grains per gram of sample, ^{Stockmarr (1971)} and ^{Maher (1981)} formula was used:

Where: Pconc is pollen per gram; R is counted pollen grains / number of spores counted; M is the number of spores added in each tablet; V is the total weight of honey sample. Pollen concentrations were classified following the scheme from ^{Maurizio (1939)}.

Botanical characterization of honey

Once the absolute number of pollen grains, honey was characterized as "mononofloral" when in its composition had a species with a percentage greater than or equal to 45% and multifloral when three or more species were with percentages ≥ 10% (Louveax and Vorwhol, 1977; Ramírez-Arriaga and Martinez-Hernández, 2009). The latter were divided into "bifloral" when two species were of importance, though it has not been formally accepted, but in this research were considered for the classification honey samples.

Statistic analysis

The results from the melisopalinological analysis of honey samples from six municipalities, were analyzed using the Statistical Analysis System (SAS) version 9.3 program using a preference of pollen analysis through PRINQUAL procedure, which performs principal component analysis (PCA) from qualitative, quantitative and mixed data, finding linear and nonlinear transformations of the variables, using partial least square, optimizing the correlation or covariance matrix properties of the transformed variables. Of all the samples taken from each municipality, recurrent species were added forming a single percentage, thus distinguishing preferences for certain species and then to the principal components analysis with qualitative data.

PROC PRINQUAL was used to perform a non-metric multidimensional preference (MDPREF) analysis. MDPREF analysis is a principal component analysis of a matrix with columns that correspond to municipalities and rows corresponding to pollen types. The percentages are the preferences of the municipalities for each pollen type. The end result of this analysis is a biplot; that is, a graphical representation of municipalities preferences for certain species.

Melisopalinological analysis

The analysis from 38 honey samples from Apis mellifera L. of Tabasco state, showed a total of 37 pollen types of importance, which were identified at genus, species and family level, the rest were assigned to morphological types (Table 2, 3, 4, 5, 6, 7).

Table 2 Results of the melisopalinological analysis of samples from the municipality of Cárdenas. 

Table 3 Results of the melisopalinological analysis of samples from the municipality of Tacotalpa. 

Table 4 Results of the melisopalinological analysis of samples from the municipality of Huimanguillo. 

Table 5 Results of the melisopalinological analysis of samples from the municipality of Centla. 

Table 6 Results of the melisopalinological analysis of samples from the municipality Centro. 

Table 7 Results of the melisopalinological analysis of samples from the municipality Balancan. 

In the municipality of Cárdenas was reported a monofloral honey sample of Cecropia obtussifolia and three from multi-floral.

For the municipality of Tacotalpa two monofloral honey samples of Diphysa carthagenensis and Mimosa albida, two bifloral samples and eight multifloral samples were reported.

For the municipality of Huimanguillo reported three multifloral honey samples.

For the municipality of Centla reported one sample of monofloral honey sample from Cocos nucifera, a bifloral sample and a multifloral sample.

For the municipality Centro reported a monofloral honey sample from mimosa albida, two bifloral samples and five multifloral samples.

For the municipality Balancan reported five monofloral honey samples, four from Rumex sp., and one from Mimosa albida, two bifloral samples and one multifloral sample.

S. mombin, (jobo roñoso), C. nucifera, B. simaruba (palo mulato), M. albida, M. pigra var. Berlandieri, M. sp., Zea mays, C. aff. diversifolia y Rumex sp., Psidium guajava, Piper sp1, sp2, sp3, Celtis, Q. oleoides, C. obtusifolia (Guarumo), H. appendiculatus (jolotzin), H. campechianum (Palo de tinto), B. verticillata, Muntingia calabura, Erythrina, D. carthagenensis, have been reported as part of the honey flora from Mexico (Espina and Ordetx 1983), Valley of Mexico (Stones and Quiroz 2007), the state of Tabasco (SAGARPA, 1998, 2002, 2003 and 2004; Cardenas, 1985; Villegas et al., 2004); Oaxaca (Santos Ramos, 2008; Navarro, 2008; ^{Ramírez-Arriaga et al., 2011}); Puebla (Ramírez-Arriaga and MartínezHernández 2007); Yucatán (Villanueva et al., 1998, 2002); Veracruz (Villegas et al., 2003); Campeche (Porter et al., 2001); South America (Sodre et al., 2007) and it is part of the bee flora by field observations in tropical regions (Espina and Ordetx, 1983).

Principal component analysis

The results from the PROC Prinqual analysis shows eigenvalues and explained variance, observing separation of the first two components explaining 65% of the variability and the rest of it (Figure 2).

Figure 2 Graphical representation of eigenvalues and explained variance. 

Since the MDPREF analysis is based on a model of principal components, the biplot, which is a graphical representation showing the relationship between the matrix data (columns and rows); the biplot dimensions are the first two components (Figure 3). The first component is one that has a greater dimension of the biplot, is the global preference of municipalities. The second component is orthogonal to the first, being the second most preferred. In the biplot, shows points that observe the pollen types and each municipality by a vector is. The points are closely grouped in a region of the graph represent the pollen types that have the same preference patterns through the municipalities. The vectors points in more or less to same direction represent the municipalities with similar preferences patterns.

Figure 3 Graphical representation of the relationship between municipalities and pollen types. 

On the upper left side can be seen the preference from M1 (Cárdenas) and M3 (Huimanguillo) by type of pollen: Cec (Cecropia obtussifolia), Bur (Bursera simaruba L.), Leg. (Leguminosae), Com. (Compositae) and Que (Quercus sp.). On the right side the preference from M2 (Tacotalpa), M5 (centro) and M6 (Balancán) by pollen type: Mia (Mimosa albida), Psi (Psidium guajava) amd Rum (Rumex sp.), and finally the preference from M4 (Centla) by Coc (Cocos nucifera) and Gra (Gramineae). This result is explained by the percentages of pollen type present in the samples taken from each municipality.