Introduction

Beekeeping is the rational exploitation of honey bees (Apis mellifera L.), for the production of honey, pollen, royal jelly, wax, and propolis, and for the use of honey bees for crop pollination¹. Mexico is one of the main producers and exporters of honey in the world. Bee-keeping is performed in all the national territory and in the country exist 41,000 beekeepers and 1.9 million colonies². During the period of 1980 to 2013, the average annual honey production was 58,182 t, with an average annual yield per hive of 29.8 kg². Mexico, is divided in five beekeeping regions: North, High-Plateau, Gulf, Pacific coast and Yucatan Peninsula³. The Yucatan peninsula region is the most important since it contributed with 34 % of the national honey production during the period of 1980 to 2013; followed by the High-Plateau region which is the second region in importance with 24 % of the national honey production during the same period².

Morelos is one of the states located in the High-Plateau region; although it only contributes with 2 % of the national honey production, beekeeping is one of main primary activities in the state, that has a long tradition among the population and that is performed through all the State, under different climatic conditions and different production systems, characteristics that place Morelos as a model for the study of beekeeping. In Morelos exist 700 beekeeping production units² and the state is a major producer of honey bee queens.

The production of honey and other honey bee products, largely depend on the ecological conditions, however their production is also affected by the social, economic and technological characteristics of the beekeepers, and by their production systems. Therefore, it is necessary to study these characteristics to understand how they influence the production processes and to generate information to support the decisions for the development of this activity. There is a wide range of methods and techniques to characterize and classify agricultural and livestock production systems, the multivariate methods, such as the principal component analysis, the factorial analysis⁴ and the cluster analysis⁵, stand out among other methods.

There are few studies about beekeeping production systems in Mexico. In a sociological study of beekeeping in the Yucatan Peninsula, the authors documented the logic of subsistence of this activity⁶. In another study⁷ that was a conducted in the state of Yucatan the socio-economic importance of the productive chain and the commercial process of honey in Mexico was analyzed; this study identified the problems of the different links of the chain using the interregional commerce theory. In the State of Jalisco⁸ a study identified five types of beekeepers, using a random sample of producers and the estimation of basic statistics.

Multivariate methods for socio-economic and technological characterization of beekeeping have not been conducted in Mexico, like those conducted in Sao Paulo⁹ and Ceará, Brazil¹.

Therefore, the objective of this study was to characterize the types of beekeepers that exist in the State of Morelos based on the use of technological components, social factors, economic factors and production factors, in order to generate information to design recommendations in support of beekeeping.

Material and methods

The study was conducted in the State of Morelos, which is located between 19° 07' 51' and 18° 19' 53" N and 99° 29' 37' and 98° 37' 59" W; the humid warm climate predominates in the state, with average annual temperature of 21.5 °C and average annual rainfall of 900 mm².

A questionnaire was designed and applied in 2012 to a sample of 116 beekeepers that was obtained from a population of 700 production units registered in the state in the honey traceability system database of the National Service for Health, Food Safety and Food Quality¹⁰. The sample size was estimated by applying the following formula for finite populations¹:

Where: n is the sample size; z is the value of Z in the table of standard normal distribution for a confidence of 95%; p is the estimator of the proportion of the characteristic investigated in the universe (p= 0.50); q is 1-p (q= 0.5); N is the number of beekeepers registered in honey traceability database and d is the sampling error (d= 0.05).

The questionnaire had two sections: I) General and socio-economic data of the beekeeper and II) Practices and technologies by area (general management, specialized management, genetics, nutrition and health), from where 16 original variables were obtained (Table 1) that were used to build six synthetic variables as technological indices for each area (Table 2).

Table 1 Social and economic variables used for the characterization and typology of the beekeepers 

Table 2 Synthetic technological variables used for the characterization and typology of the beekeepers 

The indexes were built using the methodology proposed by De Freitas and Pinheiro¹, each practice and technology took the value of 1 or 0, which indicates if the producer applied it or not. The following is the formula that was used to build the indexes:

Where Iij is the technological index of the area i for the beekeeper j, δ_in is the sum that each producer obtains based on the number of practices and technologies that he applies and δ_i...n is the maximum sum of the n practices or technologies that a beekeeper j, can perform by area i. The values of the calculated indices are within the following interval 0≥ I ij ≤1.

A total technological index IT _j was also calculated by applying the following formula:

The total index value is within the following range 0≥ IT j ≤5.

To stratify the beekeepers a multivariate analysis was performed using a factorial analysis by principal components, hierarchical clustering and K-means clustering in three steps. To select the variables, the quality, availability and relevance criteria was considered and the results of an analysis of correlation between the 22 variables included ^9,11,12. The factorial analysis by principal components without rotation was used to reduce the number of quantitative variables (Table 1), through the construction of factors that explain the greater proportion of the variance in the global analysis^9,13. The hierarchical clustering analysis, based on the Ward algorithm^14,15 to find the cut-off point in the dendrogram in graphic form, was used to identify the number of groups among the beekeepers (Figure 1), this analysis was complemented by K-means clustering^15,16 for better identification of the groups. The variables used were the factors obtained in the principal components factorial analysis that were standardized with the mean and standard deviation. Statistical analyses were performed with JMP® 4.0 (SAS Institute).

To characterize and compare the beekeeper groups, the means and standard deviation were estimated for the quantitative variables and an analysis of variance under a complete random model was performed to detect differences between groups; for the qualitative variables the frequencies were calculated and a homogeneity test was performed in order to identify differences between groups of beekeepers¹².

Figure 1 Dendrogram from beekeepers of the State of Morelos, Mexico 

Results

Typology of beekeepers

Based on the matrix of correlations of the 22 variables included (Tables 1 and 2), the 11 quantitative variables that had the highest correlations were selected, and with these variables the multivariate analysis was conducted. From the factorial analysis four factors were extracted that presented values greater than 1, based on the latent root criteria¹², these four factors explain 70.6 % of the total variation of the original variables. The factorial weight that each variable has on the extracted factors with values higher than the 0.50, were used to identify the variables associated with each factor and were used to assign an empirical interpretation and a physical name.

Factor 1 has a high correlation with the size of the production unit, and with the genetic, basic management and nutrition indexes (Table 3), so it was called productive capacity of the beekeeping production unit; it is important to notice that this new variable explains 31.4 % of the variance of the 11 studied variables, therefore the factor 1 is the factor that has more weight in the analysis, and the one that better explains the differences between the beekeepers groups.

Table 3 Factorial weights of the variables that compose the factors defined for beekeepers 

Factor 2 has a high correlation with the health index (Table 3), this index is composed by beekeeping practices to prevent and control pests and diseases, and therefore it was called health status of beekeeping production unit and explains 15.3 % of the variance.

Factor 3 shows a high correlation with the social characteristics of the beekeeper, including the age, education level and beekeeping years of experience (Table 3), these characteristics define the ability to produce of the beekeeping unit, so it was called the beekeeper capacities, this factor explains 13.9 % of variance.

Finally, factor 4 has a high correlation with the specialized management index (Table 3), which is composed of variables associated with the management of the unit, such as implementation of productive records, planning, and post-harvest handling, therefore it was called the management of the beekeeping production unit and explains 9.9 % of the variance.

The information of these four factors was incorporated to the clustering analysis to identify the groups of beekeepers; the hierarchical analysis identified three types of beekeepers in graphic form (Figure 1). This data was used to set the number of clusters in the k-means analysis; the results of the analyses indicate that there are three groups of producers and the number of beekeepers comprising each group were 64 (55 %), 10 (9 %) and 41 (35 %) for groups 1, 2 and 3 respectively.

To assign a name to each group, it was taken as reference, both, the number of hives and the technological level measured by the total technological index of the beekeeping unit. Group 1 is composed by small beekeepers with low technological level (G1), group 2 is composed by large beekeeping units with intermediate technological level (G2) and group 3 are medium size beekeepers with intermediate technological level (G3) (Table 4).

Table 4 Mean ± standard deviation of the variables used for the characterization of the beekeepers 

^abc Values with different supercript differ (P<0.05).

Discussion

The four factors obtained from the factorial analysis explains 71 % of the existing variation between the production units included in the study. This value is considered as acceptable taking into account that in the social sciences is normal to consider solutions that represent 60 % of the total variance¹⁷ and it is higher than the value reported in Argentina, of 68 % when agricultural producers were typified¹⁸.

The classification of producers in three categories: small, medium and large, generated by the size of the production unit (number of hives) in this study coincides with the classifications of beekeepers in the State of São Paulo, Brazil, which mention that small beekeepers have 10 to 50 hives, medium beekeepers from 51 to 200 hives and the large beekeepers more than 200 hives⁹. However, the classification criteria in two levels, low and intermediate, obtained for the technological level in this study, differs from the classification generated in another study of beekeepers in Mexico⁸, in that study three types of beekeepers were identified: technified, semi-technified and traditional; but our classification coincides with was reported in the State of Ceará, Brazil where beekeepers with low and medium technological level where found¹.

The productive capacity of a company is determined by the stock translated in machinery and equipment, by the capabilities of the staff and the technology used²⁰. In the case of the beekeeping activity, the investment in machinery and equipment is minimal, so the beehives represent the main investment, so its value is the stock of the beekeeper, therefore the production depends on the number of hives and the technology used in the areas of colony management, nutrition and genetics.

Analyzing the characteristics of the variables included in Factor 1, three of the variables are related to the size of the production unit; this result coincides with a study in Switzerland²¹, where the size of the production unit, measured by the number of colonies was the most important factor that affects honey production.

The use of management practices in beekeeping operations has been evaluated in the State of Ceará, Brazil for sedentary and migratory beekeeping¹; in this study the authors found that the adoption index was 0.59 and 0.61 respectively, these values are smaller than the one found in this study for the three groups of beekeepers identified in the state of Morelos, México.

Feeding honey bee colonies was evaluated in production units of the humid tropic of Mexico²²; the authors report an adoption index of 0.52, this value is similar to index of 0.52 of the small beekeepers with low technological level (G1) identified in this study, but it is lower than the index of 0.81 of the large beekeepers with intermediate technology level (G2) and the index of 0.93 of the medium beekeepers with intermediate technology level (G3) of 0.93, these results are an indicator of the importance that the producers give to this practice in Morelos.

The presence of honeybee diseases and pests is one of the main factors affecting the production of honey and other bee products, the three groups of beekeepers had low values for Factor 2; this differs from what was reported in the humid tropics in Mexico²², where activities related to pests and diseases control were the most important for the producers.

The beekeeper is the human stock of the beekeeping production unit, and is considered as a facilitator of economic growth and development. The averages for the variables that make up the Factor 3 were similar for the three groups of beekeepers. Several authors^9,22,23 agree on the importance that the age, education level and years of experience in beekeeping as elements that favor or impede the use of innovations. Finding no difference among the three groups of beekeepers for these variables, coincides with what is reported for beekeepers of Uganda²⁴.

The average age that the beekeepers of the three groups presented corresponds to the adult stage, which influences the use and adoption of innovations, and it is also an important factor that affects the administrative and technical management of the unit⁸. The education level of the beekeepers of groups G1 and G2 correspond to high school education (9 yr) and the education level of the beekeepers of the group G1 corresponds to elementary education and one year of high school (7 yr), these results are similar to those reported in other studies^21,22,24, but differs from what is reported in Saudi Arabia²⁵, where it was found that 40.7 % of beekeepers had college education level. The experience in the activity represented by the years of being a beekeeper, was higher in the group G2 with an average of 22 yr, followed by group G3 with 19 yr and the beekeepers of group G1 had an average of 15.5 yr. These results coincide with those reported by other authors^8,22,25, and differ from the results in Saudi Arabia and Nigeria^24,26, that reported beekeepers with six or fewer years of experience in beekeeping. Although the experience in the practice of the activity has no relationship with the adoption of technology²⁴ it is positively related to honey production²¹.

The activities related to the Factor 4 involve a greater investment of resources, both economic and human; these activities were used to estimate the index of specialized management, in which there were no differences between groups, however in 6 of the 12 activities that make up the index, differences between the groups of beekeepers were found.

Medium beekeepers with intermediate technology level (G3) mobilize colonies for the production of honey, while large beekeepers with intermediate technology level (G2) do it less frequently and small beekeepers with low technological level (G1) do not mobilize colonies. This can partly explain the higher productivity measured as the average annual honey yield per colony of beekeepers of group G2 in comparison with the other two groups, mobilizing colonies allows beekeepers to have a higher number of harvests of honey during the year.

The producers of the groups G2 and G3 produce beeswax foundation and replace old combs; the beekeepers of group G1 don´t do these activities, this is consistent with results from other studies that compared the technological level between producers that mobilize colonies and producers that don´t mobilize colonies^1,9.

Beekeepers of the group G1 grow honey producing plants, while producers of the other two groups don´t, this is due mainly because G1 beekeepers do not mobilize colonies for honey production and these beekeepers try to ensure a source of nectar for their colonies, similar results were reported in Nepal¹⁹.

The proportion of beekeepers that produce royal jelly and propolis was higher in groups G2 and G3 respectively, while in the group G1 no beekeepers were found that obtain any of these two products. However the proportion of beekeepers producing royal jelly of the group G2 was low (18 %), as well as the proportion of beekeepers producing propolis of the group G3 (7 %).

Finally the group of medium-sized beekeepers with intermediate technology level (G3) keep productive records more frequently than the beekeepers of the other two groups. Forty nine percent (49 %) of the producers of the G2 group were involved in this activity, which may explain in part the higher productivity measured as the average annual honey production of this group.

Conclusions and implications

The use of multivariate analysis allowed to identify three types of beekeepers in Morelos, small with low technological level (55 %), large with intermediate technology level (9 %) and medium with intermediate technology level (35 %). The variables that were relevant for its stratification were those related to the productive capacity of the beekeeping production unit, such as the size of the production unit (number of hives), the use of technological components for colony management, genetics, nutrition, health and the management of the production unit. The small beekeepers with low technological level use less technology in colony management, nutrition and genetics than the medium and large beekeepers with intermediate technology level, which has an impact on productivity. There is potential to improve the productivity of the beekeeping units in Morelos, designing policies that promote the use of technology in all three types of producers.