Typology of dairy production systems in the Eastern Amazon, Pará, Brazil

V V Dantas, R P Oaigen, M A S Santos¹, C S S Marques and F Silva

State University of Pará (UFPA).
Nucleus of Agricultural Science and Rural Development (NCADR) Belém - Pará - Brazil
vd_dantas@hotmail.com
¹ Federal Rural University of the Amazon (UFRA).
Postal Code: 66.077-530. PO Box 917. Belém, Pará, Brazil.

Abstract

This study aimed to characterize dairy production systems in the Eastern Amazon state of Pará, Brazil. The data were collected using questionnaires to 112 dairy farmers located in the southeastern and northeastern mesoregions of the state.

A typology was established using a Factor Analysis (FA) method that identified four factors accounting for 66.99% of the variance in the original variables and a cluster analysis (CA) that identified four groups of farmers.

Groups I and II were characterized as family farms with low skill levels and low productivity. Group III corresponded to farmers that exhibited greater technology adoption, including feed supplementation, health management, milking technology and activity management, consequently inducing better performance, who achieved productivity levels of 5.05 (L/cow/ day). Group IV consisted of family farmers who produce milk and grow agricultural products at smaller scales and with lower economic returns.

This study emphasized that farmer education level indicates greater potential for technology innovation adoption. In general, production systems with higher-level management are more competitive and achieve better results compared to those that do not adopt technologies.

Keywords: dairy cattle farming, family farming, technology

Introduction

Milk production is one of the most important components of Brazilian agriculture, accounting for a significant portion of the revenues derived from agribusiness, approximately 19% of the Total Gross Value of Production (GVP). This industry has undergone several structural transformations in recent years, which have induced changes to the competitive environment of this agro-industrial sector (Barros et al 2010).

Brazilian milk production systems are heterogeneous and consist of a large number of farmers, from unskilled to technologically adept, who establish production units with different levels of technology adoption, productivity and income (Leite and Gomes 2001).

Dairy farming operates in small production units and exhibits seasonality in the Eastern Amazon. The abundance of pasture during the rainy season provides ample and quality feed and, therefore, leads to higher production. Conversely, milk production decreases drastically during the dry season (off season) due to the shortage of pasture, turning farmers into temporary producers who concentrate their production during the rainy season (Santana 2002; Martins et al 2008).

This production variability requires the introduction of different actions that meet the needs of various systems and farmer groups (Aleixo et al 2007). It is important in this context to correctly characterize production systems to identify technological structures and to begin to enhance and promote livestock farming through agricultural extension and technical assistance programs; this can be done by studying production units and by guiding decision-making about further research (Moura et al 2013).

The typology of a production system may be established according to a variety of analyzed resources, including the herd size, milk production area (land available for grazing), milking practices, milk cooling procedures, feed management, herd health management, herd reproductive management and the general management practices (Tsioulpas et al 2007; Bodenmüller Filho et al 2010).

However, characterizing dairy production systems is complicated due to the lack of standardized assessment methods (Pereira 2001). The representativeness of individual characteristics or variables is limited in global assessments because the numerous characteristics are redundant as a result of correlations; when a characteristic is a linear combination of others, further complicating data analysis and interpretation (Liberato et al 1999).

Accordingly, the multivariate analysis of production systems, especially factor analysis (FA) combined with cluster analysis (CA), is useful for descriptive analysis and the simultaneous interpretation of data from several quantitative and qualitative perspectives; this provides the opportunity to refine characterization studies of dairy production systems (Aleixo et al 2007; Carrillo et al 2011). These methods enable specific and precise evaluation of complex phenomena because they treat several variables simultaneously, even when a theoretical model of the relationships between the variables is unknown (Hair Junior et al 2005; Bakke et al 2008).

Research related to production system typologies is very important for categorizing farmer groups, thereby enabling the identification of opportunities and barriers to adoption, or the outright rejection of technologies; in turn, this supports the development of specific public policies and agricultural extension programs that take into consideration the local and regional reality for farmers.

Therefore, this study aimed to characterize milk production systems in the Eastern Amazon, state of Pará, Brazil using farmer grouping based on common characteristics, and a subsequent evaluation of related technical and economic variables.

Materials and methods

Study area

This study was conducted in the state of Para, which covers most of Eastern Amazon of the Brazil. The mesoregions Southeastern and Northeastern of Pará were chosen to characterize the production systems because together they totalize 86.6% of milk production and encompass major ecosystems of dairy farming. (Figure 1).

Figure 1: Study area location

The southeastern mesoregion of the Pará has a total area of  297,344.257 km² and a population of 1,719.989 inhabitants. Presents as economic base the agricultural activity with a bovine population of 12,254.159 heads with aptitude to cut and milk and milk production is an alternative widely adopted in family farms (IBGE 2010; IBGE 2012).

The northeastern mesoregion of the Para covers an area of 83,074.047 km² with a population of 1,903.264 inhabitants. The economy is based on agriculture with 1,259.273 heads of cattle, the extraction and processing of wood, but its strongest feature is the subsistence agriculture (IBGE, 2010; IBGE, 2012).

Data collection

Data were collected from interviews with 112 dairy farmers between April and May, 2013. The calculation used to assess the sample size was performed using data from the Municipal Departments of Agriculture, the Technical Assistance and Rural Extension Company of the State of Pará - EMATER-PA) and the Agricultural Protection Agency of the State of Pará - ADEPARÁ) using equation 1:

where: n = calculated sample, N = number of farmers, Z = confidence level, e = sampling error, p = percentage by which the phenomenon occurs and q = additional percentage (100-p). A 95% confidence interval and an error with a 5% significance level were adopted. The total sample size was 110, although 112 questionnaires were undertaken due to farmer availability.

A four-part structured questionnaire titled “drivers of competitiveness” was prepared and validated based on suggestions from a team of professors, researchers and specialists from the fields of agricultural and social sciences. The drivers were defined as follows: the production system and farmer socioeconomic profile; milking hygiene and sanitary conditions; activity management; and the external environment. Each driver consisted of a group of related questions. The interviews consisted of a list of question and response options; the sequence of questions was identical for all respondents to ensure that variations between the responses resulted from individual differences and not from the interviewers’ approach. The degree of importance of each driver within the system was assessed based on the responses when analyzed as the main factors that affect dairy farming performance and, therefore, its competitiveness.

Multivariate data analysis

Factor Analysis

FA, which is a multivariate statistical method used to reduce and summarize data, was applied to examine the relationships between the original variables (Hair Junior et al 2005; Härdle and Simar 2007).

Therefore, the objective is to identify factors that are not directly observable from a correlation between a set of observable and measurable variables (Bezerra 2007). The basic model of FA is as follows:

X_i = α_i1F₁ + α_i2F₂ + α_i3F₃ + ... + α_ijF_j + e_i

Equation 2

wherein, X_i represents the standardized variables; α_i represents the factor loadings, F_j represents the unrelated common factors and e_i represents an error that is the percentage of variation of variable i that is exclusive to it and is not explained by a factor or another variable from the analyzed set.

Bartlett's sphericity test and a Kaiser-Meyer-Okin (KMO) test were performed to assess the appropriateness of applying a FA The former tests the hypothesis that the correlation matrix is ​​an identity matrix; that is, its determinant is 1 and all of the other values are 0. This property indicates that no correlation occurs between the variables. The KMO is used to assess the input value of the variables in the model, which may range from 0 to 1 (Mingoti 2005). Values of 0.80 or higher are considered optimal; 0.70-0.79 are considered moderately good; 0.60-0.69  are considered poor; and 0.50-0.59 are considered very poor. The Categorization of these values enables the use of an FA. Values lower than 0.50 are considered unacceptable for this type of multivariate analysis (Pereira 2001).

The factors were estimated using SPSS^@ statistical software (version 18.0) by applying a principal component analysis; those components with characteristic roots higher than 1 were extracted. The factor scores were estimated based on these factors and were used in a CA. The dummy variables used in the factor analysis were as follows: V1 (infrastructure), V2 (milking management), V3 (herd feed), V5 (asset control), V6 (herd identification), V7 (access to technological information), V8 (farmer-supplier relationship), V9 (farmer-dairy products relationship), V10 (herd health), V11 (herd reproduction), V12 (animal welfare), V13 (Organization of producers) and V14 (credit policies and sanitary inspection). It is noteworthy that the variables were calculated according to the number of positive responses and were transformed into percentages to assess the value used in the analysis because the data are qualitative.

Cluster Analysis

 A CA was used to group elements based on their similarity within a group and the heterogeneity between them (Kaufman and Rousseau 1990). CA is used to organize or classify several individuals into a small number of groups, often termed taxonomies or typologies and may be performed using different methods, including hierarchical methods. The partitioning in this method begins with an undefined group number, wherein the main groups are divided into minor subgroups, involving the clustering of individuals who have similar characteristics. The classification of individuals into different groups is performed using a clustering function (distance or similarity) and a mathematical clustering criterion; the Euclidean distance is a dissimilarity measurement that is most commonly used. Interaction between the K groups occurs in a non-hierarchical way, using certain criteria to minimize intergroup variance, which may have a negative impact on the quality of the generated clusters (Pohlmann 2007).

The main advantage of hierarchical algorithms is that they provide not only clustered results but also data structure, and they facilitate the generation of sub-clusters from these data. Furthermore, they enable direct visualization with a dendrogram to show how the data are linked and to identify the actual similarity between two different points belonging to the same cluster or to a different one (Linden 2009).

Results

Characterization of Technological Factors

Four factors with characteristic roots higher than 1 were extracted, which explained 66.99% of the total data variance. The Bartlett’s test was significant at 1% probability, rejecting the null hypothesis that the correlation matrix was an identity matrix. The KMO test exhibited a value of 0.839, indicating that the data sample was adequate for an FA (Table 1).

The first factor explained 28.12% of the total data variance and was strongly and positively associated with infrastructure, milking management, herd feed, activity management, asset control and herd identification variables. The combination of these variables into a single factor may be explained by the fact that these technologies exhibited the highest level of adoption among farmers. The exceptions included the management and assets control indices, which exhibited very low representativeness within the factor and were often used simultaneously. This factor was termed “infrastructure, milking, management and feed management technology” (Figure 2). Farmers positioned positively on the X-axis (F2) and Y-axis (F1) had a greater association with the use of these technologies, based on a Cartesian coordinate system.

The second factor represented 14.94% of the original variability and was positively correlated with the following variables: access to technological information, farmer-supplier relationships and farmer-dairy product relationships. The association of these variables usually characterizes a farmer’s interest in new technological information and their external relationships with suppliers and buyers; it can be termed “the search for innovations and market relationships”. The degree of farmer fidelity (commitment to repurchase or to providing a product or service) to companies supplying inputs was, strikingly, only 50%. Conversely, the percentage regarding farmer fidelity to dairy products was higher, at 83.9%.

The third factor accounted for 14.21% of the total data variance and exhibited a positive association with herd health management, reproduction management and animal welfare. The combination of these variables can be summarized as “Production management technologies”. Farms with high scores in this factor exhibited better reproductive management and, therefore, better animal husbandry indices. However, the level of technology adoption in the study area is still considered low because only 0.9% of farmers use artificial insemination. This limited application may be related to a lack of technical knowledge and, in other cases, to method failure among farmers resulting from heat stress, poor cattle nutrition and inefficient detection of external manifestations of the estrous cycle in cows, eventually limiting the success and expanded implementation of artificial insemination.

The fourth factor represents 9.72% of the original data variance and exhibited a positive relationship with farm organization membership, credit policies and sanitary inspection variables. The association of these variables may be characterized globally as “Rural organization and credit policies”. Farm organization membership is essential for achieving the best prices for the product, transforming individual investments into collective investments and channeling more investment into dairy farming (Figure 3).

Establishment of a typology

A CA was performed based on the scores of the four factor estimates. An analysis of the results included hierarchical clustering into 4 groups, wherein production systems with similar characteristics were clustered using a squared Euclidian distance.

The results indicated that the main limiting factors of farm viability were low milking productivity and low levels of skilled labor, which translated into the inefficient use of production structures and technology. Groups I, II and IV represented the lowest production strata, with mean areas ranging from 99.7 ha to 161.1 ha, daily production from 65  to 100 L day ^-1  and productivity from 4.23 to 4.34 (L cow ^-1 day ^-1). Group III included farmers exhibiting greater technology adoption and better production performance than the others, with mean area, milk production and productivity of 363 ha, 213 (L day ^-1) and 5.04 (L cow^-1 day^-1), respectively. However, these values are much lower than in other Brazilian regions. The general characteristics of the groups are outlined in Table 2.

Group I – Family dairies with intermediate levels of expertise

Group I comprised 17 farmers, which corresponded to 15.17% of the sample dairy farms. Most farmers were older than 50 and had a low level of education; 70.6% had incomplete primary education, and 17.6% were illiterate. This group exhibited one of the lowest income levels from dairy production. This group exhibited a low level of dairy farming skills, despite having more farming experience (over 10 years). The farms exhibited mean milk production of 100 liters per day because they had the second largest average cow herd in the area (mean of 57 animals) and, notably, this group exhibited the second lowest mean (4.24 L cow^-1 day ^-1) productivity index.

This group exhibited the second largest dairy production area, with more than half assigned for pasture production, although inadequate management contributed to pasture degradation, accounting for poor nutrition. The low use of herd feed supplementation was associated with poor nutrition, which led to low productivity. 94.1% of the farmers grow maize and cassava using available family labor; however, dairy farming is still considered to be their main activity.

The use of technology and management practices in animal production and the access to technology information index were deficient because all of the farmers from this group disclosed that they have no access to technology information and transfer. This group exhibited the highest mean use of rural credit, although most farmers stated that they use most of their funds for other purposes, including purchasing cars and household appliances, when asked whether the funds were used to conduct farming activities.

Group II – Farmers with family dairies and low skill levels

Group II consisted of 19 farmers (16.96% farms) with less experience in dairy farming and lower education levels; this group had the highest percentage of illiterate farmers (31.6%) and the greatest number of farmers younger than 35 years. Of the farmers in this cluster, 78.9% exhibited incomes ranging from 1 to 5   the minimum wage, which is a result of families receiving supplementary income from government assistance programs, including a family allowance, and many farmers conducting other activities, including working in service industries (tractor driving and masonry, among others); however, the farmers in this group claimed that milk production was the main source of their family income.

This group was similar to group I, in that they were low-income dairy farms. This group had the lowest mean productivity of 4.23 (L cow^-1 day^-1), despite having the second highest mean of feed supplementation adoption (53.2%), assessed by farmers supplying concentrates. Most of the farm area was dedicated to pasture production and crops, including rice, maize and cassava. The farm work was conducted by family members, which reinforces the characteristics of family dairy farming in this region. The degree of dairy management skills was low. However, this group had the highest mean of farmer access to technological information, which is explained by the fact that most lived near urban centers, with greater access to newsletters and information gathered from vendors.

Group III – Family farmers with the highest level of dairy farming skills

 Group III consisted of 11 farmers. They exhibited the highest level of education, some with secondary education and others with higher education. They were mostly concentrated in the 35-45-year-old age group, and may be considered relatively young farmers. A considerable portion of these farmers had noticeably little experience, with less than 5 years of farming. Dairy farming was considered to be the main activity for 82% of these farmers. The income from production in this group is notable because they had the highest mean gross income, which may indicate a higher level of dairy farming skill. This group also had the largest area dedicated to dairy farming, the largest cow herds and the highest mean productivity, producing an average of 213.77 (L day ^-1). Family labor was also present, albeit to a lesser extent, with a higher index of hired labor compared to the other clusters.

The use of technologies specific to the development of dairy farming was relatively widespread, including the use of feed supplementation, dairy management methods and quality control (namely cooling, proper hygiene and sanitary management using vaccinations). Furthermore, the group also exhibited the greatest use of agricultural management methods. This group had the highest mean of technical assistance, which indicates that training services may contribute to dairy farming success. This group received the least amount of government incentives to fund their dairy farming costs. This low level was by farmer choice, and they consequently have greater financial and production capacity for investments in production systems.

Group IV – Diversified family farming with low skill levels

Group IV consists of 65 farmers, corresponding to 58.03% of the farms. It indicates similar characteristics to groups I and II. Of these farmers, 72.3% had primary education, and the group had intermediate experience in dairy farming. This cluster noticeably had 32.3% farmers receiving less than minimum wage. Furthermore, they had the highest percentage of farms without access to electricity. This is the group with the smallest production area (a mean of 99.7 ha), smallest herd size (mean of 32.9 cows/year) and lowest production level (mean of 65.9 L /day). They exhibit diverse systems, incorporating agriculture (maize, rice and cassava) and pig farming, although dairy farming was considered the main activity. This group exhibited the lowest level of technology applied to herd feed of all of the groups, which indicates an exclusively pasture-based system with a low level of feed supplementation.

This group had the highest ratio of family labor, which is common among the lowest production strata. Furthermore, they exhibited the lowest index of technology applied to dairy management and infrastructure. This group exhibited the second lowest index of technological information in terms of their access to services because this group received the least technical assistance. This group exhibited the second lowest mean value of farmers benefiting from dairy farming subsidies.

Discussion

Several studies have been conducted using CAs to characterize milk production systems, according to Vargas-Leitón et al (2013). The number of clusters generated in these studies usually ranges from 3 to 5. The typologies are not directly comparable because the number and type of variables differ considerably, although the clustering method is similar.

This study identified similarities between dairy production system production and socioeconomic characteristics, corroborating results from other studies. Assessing the smallholder farmer socioeconomic and production characteristics in central Mexico, Martinéz-García et al (2012) identified four groups of small-scale dairy farms that were mainly differentiated by the farm area, dairy herd size, milk production, herd management, use of technology, level of education, source of income and government subsidies and financial aid programs. Bernués and Herrero (2008) identified three production systems when studying technology adoption is dairy systems with mixed cultures in Santa Cruz, Bolivia; these were mainly differentiated by production, socioeconomic and technological characteristics. Characterizing dairy herds in Costa Rica, Vargas-Leitón et al (2013) identified 5 different groups, encompassing variables related to physical characteristics, management and production levels. In a study of milk production in the central highlands of Mexico, Espinoza-Ortega et al (2007) found three different small-scale groups that were differentiated by farm size, use of technologies, herd size, production management, feed and access to technical assistance.

The 4 groups identified for this study were primarily distinguished by the farm area size, dairy herd size, milk production, education level, production management and activity management. Characteristics, including education levels and mean age of farmers, indicated that those characteristics are important predictors of farmer adaption to recent agro-industrial milk market dynamics, considering their resistance to innovations. Espinoza-Ortega et al (2007), Bernués and Herrero (2008) argued that the decision to accept or reject technology is attributed farmer heterogeneity, especially characteristics related to age and education. This phenomenon was observed in this study, where simplicity and resistance to innovation are striking characteristics of the interviewed farmers, who exhibited greater difficulty adapting to new challenges. Furthermore, they did not feel motivated and believed that technological innovations added little to their operation and that their acquired experience was meeting their needs.

Group I, whose farmers had been involved in the activity for more than 10 years, noticeably exhibited lower performance than group III, whose farmers had less than 5 years of experience. The farmer’s experience becomes a positive factor because it may contribute to a successful enterprise (Avilez et al 2010; Martinéz-Garcia et al 2012). However, other characteristics must be considered concurrently, including farmer capacity to gather and process information, their ability to use agricultural technologies and their predisposition to new challenges. The level of education was a good indicator of this capacity (Solano et al 2000).

Therefore, a farmer’s length of experience (measured by their age or work years) individually contributes little to performance because older farmers may show less stamina, poorer management practices, a lesser understanding of planning methods and an increased resistance to the adaptation to new changes required by the market. Conversely, younger farmers are more attracted to innovation and may exhibit a greater willingness to seek knowledge, an ability to adjust to transformation in the sector and better performance; therefore, this leads to the greater competitiveness that was observed in this study. Parré et al (2011) also argued that younger farmers that are new to the field have not experienced a variety of organizational and competitive issues, which gives them greater capacity for adaptation to recent changes in the sector when combined with other characteristics, including their level of education.

Group III exhibited the highest productivity index, above the national and state means of 4 and 2 (L cow^-1 day^-1), respectively. Santos et al (2011) and Sena et al (2012) identified a mean productivity of 3.78 (L cow^-1 day^-1) when conducting studies about dairy production in the states of Rondônia and Pará. Moura et al (2010) identified three groups of farmers when characterizing dairy production systems in the Cariri region of Paraíba, and the group with the lowest mean productivity exhibited a value of 9.83 (L cow^-1 day^-1). Aleixo et al (2007) identified that the group with the lowest mean productivity exhibited a value of 8.22 (L cow^-1 day^-1) during the rainy season and 7.50 (L cow^-1 day^-1) during the dry season, when characterizing the National Agroindustrial Cooperative of farmers (Cooperativa Nacional Agro Industrial, COONAI) in the state of São Paulo. Lopes et al (2007) identified a mean productivity of 9.86 (L cow^-1 day^-1) in the state of Goiás, 12.32 in Minas Gerais, 18.91 in Paraná, and 18.77 (L cow^-1 day^-1) in the state of Rio Grande do Sul, in a study of production costs and the scale of dairy farming in the main Brazilian states.

These data prove the low productivity of dairy production systems in the state of Pará compared to other Brazilian regions, and they show the need to propose alternatives to increasing production and productivity because the scale of production has a great effect on dairy farming profits. These approaches, directed towards herd genetic breeding, feed, health and adoption of methods aimed at animal welfare, stand out among the alternatives to improving production indices.

All of the clusters exhibited limitations in terms of farmer access to technological information. Consequently, assistance becomes a key factor in spreading knowledge among farmers, contributing to the performance of dairy farming. However, Batalha et al (2005) argued that this information and technology often failed to turn into innovation, given the farmers’ lack of capacity and the conditions for innovation, even when access to them was available. Performance and viability depend on a set of factors and require a systemic approach, with an emphasis on resource availability, technical guidance and improved perspectives.

The group III farmers also exhibited a higher index of technology adoption, albeit lower than farms from other Brazilian regions. However, the ranking of factors in this group clarified that the level of technology adoption related to feed and health management was higher than the use of technology management and milking mechanization. Kiptot et al (2006) claimed that farmers were more concerned about adopting easily applicable or operational methods with immediate benefits over more complex, systematic and/or strategic methods. However, the adoption of other methods, including management practices, is of the utmost importance for dairy farming success, given the transformations that has occurred in the dairy sector in recent years.

Conclusions

• Field management provides competitiveness gains for farmers, enabling them to reach their optimal profitability, plan for the future, define investments and correct actions. Furthermore, management practices help rural farmers in decision-making using historical data. The application of specific public policies that support farmer training are needed for this purpose because of the diverse levels of education among farmers.

  
  

• Dairy farmers in southeast and northeast Pará were clustered into four groups related to socioeconomic and production variables, according to the CA. Their educational level indicates a greater possibility for the introduction of new technologies in dairy farming, whereas the presence of management tools on the property predicts higher production and profitability.

  
  

• The dairy production systems that comprised the four study groups were fragile, with low levels of production and productivity. The implementation of public policies that support enhancing the development of dairy farming by extending rural credit and expanding assistance and technician training based on farmer needs are important to stimulate technology adoption. However, it is important to emphasize that an effective process of technology adoption and, consequently, economic efficiency improvements in production systems must comply with the specific and intrinsic characteristics of different farmer groups.

References

Aleixo S S, Souza J G e Ferraudo A S 2007 Técnicas de análise multivariada na determinação de grupos homogêneos de produtores de leite . Revista Brasileira de Zootecnia, 36 (6): 2168-2175. (supl.) Retrieved november 16, 2013, from: http://www.scielo.br/pdf/rbz/v36n6s0/29.pdf  

Avilez J P, Escobar P, Fabeck G, Villagran K, García F, Matamoros R y Martinez Garcia A 2010 Caracterización productiva de explotaciones lecheras empleando metodología de análisis multivariado. Revista Cientifica (Maracaibo), 20 (1): 74-80. Retrieved February 04, 2014, from: http://www.uco.es/zootecniaygestion/img/pictorex/10_10_53_pdf_6.pdf  

Bakke H A, Leite A S M e Silva L B 2008 Estatística multivariada: Aplicação da análise fatorial na engenharia de produção. Revista Gestão Industrial, 4 (4): 01-14. 

Barros F L A, Lim J R F e Fernandes R A S 2010 Análise da estrutura de mercado na cadeia produtiva do leite no período de 1998 to 2008. Revista de Economia e Agronegócio , 8 (2): 177-198. Retrieved April 23, 2012, from: http://www.revistarea.ufv.br/index.php/rea/article/view/165/173   

Batalha M O, Buainain A M e Filho H M de S 2005 Tecnologia de Gestão e Agricultura Familiar. In: Batalha M O, Filho H M de S (org). Gestão Integrada da Agricultura Familiar. São Carlos. EdUFSCar. . Retrieved Junuary 04, 2014, from: http://www.sober.org.br/palestra/12/02O122.pdf  

Bernués A and Herrero M 2008 Farm intensification and drivers of technology adoption in mixed dairy–crop systems in Santa Cruz, Bolivia. Journal of Agricultural Research 6 (2): 279–293. Retrieved December 12, 2013, from: http://digital.csic.es/bitstream/10261/10354/1/HerreroM_SpJAgrRes_2008.pdf  

Bezerra F A 2007 Análise fatorial. In. Corrar L J, Paulo E, Dias-Filho J M (Coord.) Análise multivariada: para os cursos de administração, ciências contábeis e economia. São Paulo: Atlas. p.73-130. 

Bodenmüller Filho A, Damasceno J C, Previdelli I T S, Santana R G, Ramos C E C O e Santos G T 2010 Tipologia de sistemas de produção baseada nas características do leite. Revista Brasileira de Zootecnia, 39 (8): 1832-1839. Retrieved December 18, 2013, from: http://www.scielo.br/pdf/rbz/v39n8/v39n8a28.pdf  

Carrillo L B, Moreira L V H y Gonzalez V J 2011 Caracterización y tipificación de sistemas productivos de leche en la zona centro-sur de Chile: un análisis multivariable. Idesia, Arica, 29 (1): 71-81. Retrieved December 12, 2013, from: http://www.scielo.cl/pdf/idesia/v29n1/art10.pdf  

Espinoza-Ortega A, Espinosa-Ayala E, Bastida-López J, Castañeda-Martínez T and Arriaga-Jordán C M 2007 Small-scale dairy farming in the highlands of central Mexico: technical, economic and social aspects and their impact on poverty. Experimental Agriculture 43 (2): 241–256.

Hair Junior F, Babin J B, Anderson R E, Tatham R L and Black W C 2005 Multivariate data analysis. 6.ed. Upper Saddle River: Prentice Hall. 928p. 

Härdle W and Simar L 2007. Applied multivariate statistical analysis. 2. ed. Berlin: Springer. 

IBGE – Brazilian Institute of Geography and Statistics 2010 Produção da Pecuária Municipal, 38: 1-65. Retrieved june 22, 2013, from: http://biblioteca.ibge.gov.br/visualizacao/periodicos/84/ppm_2010_v38_br.pdf  

IBGE – Brazilian Institute of Geography and Statistics 2012 Produção da Pecuária Municipal, 40: 1-65. Retrieved august 22, 2013, from: ftp://ftp.ibge.gov.br/Producao_Pecuaria/Producao_da_Pecuaria_Municipal/2012/ppm2012.pdf

Kaufman N L and Rousseau W 1990 Finding groups in data: an introduction to cluster analysis. New York: John Wiley & Son, 368 p. 

Kiptot E, Franzel S, Hebinck P and Richards P 2006 Sharing seeds and knowledge: farmer to farmer dissemination of agroforestry technologies in western Kenya. Agroforestry Systems, 68 (3): 167–179. Retrieved august 12, 2013, from: http://www.worldagroforestry.org/downloads/Publications/PDFS/ja06129.pdf  

Leite J L B e Gomes A T 2001 Perspectivas futuras dos sistemas de produção de leite no Brasil. In: Gomes A T, Leite J L B, Carneiro A V (Eds.). O agronegócio do leite no Brasil. Juiz de fora: Embrapa/CNPGL, P. 207-240. 

Liberato J R, Vale F X R e Cruz C D 1999 Técnicas estatísticas de análise multivariada e a necessidade de o fitopatologista conhecê-las. Fitopatologia Brasileira, 24: 5-8. 

Linden R 2009 Técnicas de agrupamento. Revista de Sistemas de Informação da FSMA, 4: 18-36. Retrieved march 02, 2013, from: http://www.fsma.edu.br/si/edicao4/FSMA_SI_2009_2_Tutorial.pdf   

Lopes P F, Reis R P e Yamaguchi L C T 2007 Custos e escala de produção na pecuária leiteira: estudo nos principais estados produtores do Brasil. Revista de Economia e Sociologia Rural, 45 (3): 567-590. Retrieved june 22, 2013, from: http://www.scielo.br/pdf/resr/v45n3/a02v45n3.pdf  

Martinéz-García C G, Dorward P and Rehman T 2012 Farm and socio-economic characteristics of smallholder milk farmers and their influence on technology adoption in Central Mexico. Tropical Animal Health Production, 44 (6):1199–1211. 

Martins G C C, Rebello F K e Santana A C 2008 Mercado e dinâmica espacial da cadeia produtiva do leite na região Norte. Belém, Banco da Amazônia. (Estudos Setoriais, 6), 50 p. 

Mingoti S A 2005 Análise de dados através de métodos de estatística multivariada: uma abordagem aplicada. Belo Horizonte: Editora UFMG. 

Moura J F P, Pimenta Filho E C, Gonzaga Neto S, Cândido E P, Menezes M P C, Leite S V F e Guilhermino M M 2010 Caracterização dos sistemas de produção de leite bovino no Cariri paraibano Maringá, 32 (3): 293-298. Retrieved August 14, 2012, from: http://periodicos.uem.br/ojs/index.php/ActaSciAnimSci/article/view/7123/7123   

Moura J F P, Pimenta Filho E C, Gonzaga Neto S e Cândido E P 2013 Avaliação tecnológica dos sistemas de produção de leite bovino no Cariri da Paraíba. Revista Brasileira de Saúde e Produção Animal, 14 (1): 121-131. Retrieved October 30, 2013, from: http://www.scielo.br/pdf/rbspa/v14n1/13.pdf  

Parré J L, Bánkuti  S M S e Zanmaria  N A 2011 Perfil socioeconômico de produtores de leite da região sudoeste do Paraná: Um estudo a partir dos diferentes níveis de produtividade. Revista de Economia e Agronegócio, 9 (2): 275-300. Retrieved april 22, 2013, from:  http://ageconsearch.umn.edu/bitstream/121299/2/Artigo%206.pdf  

Pereira J C R 2001 Análise de Dados Qualitativos – Estratégias Mercadológicas para as Ciências da Saúde, Humanas e Sociais. 3ª ed. – São Paulo: Editora da Universidade de São Paulo. 

Pohlmann M C 2007 Análise de conglomerados. In. Corrar, L.J., Paulo, E., Dias-Filho, J.M.(Coord.). Análise multivariada: para os cursos de administração, ciências contábeis e economia. São Paulo: Atlas, p. 73-130. 

Santana A C 2002 Análise da comercialização e dos custos na cadeia produtiva de leite na Amazônia. In: Santana, A.C., Amin, M.M. Cadeias produtivas e oportunidades de negócios na Amazônia. Belém: UNAMA, 454 p. 

Santos M A S dos, Santana A C e Raiol L C B 2011 Índice de modernização da pecuária leiteira no estado de Rondônia: determinantes e hierarquização. Perspectiva Econômica, 7 (2): 93-106. Retrieved december 20, 2013, from: http://revistas.unisinos.br/index.php/perspectiva_economica/article/view/pe.2011.72.03/645   

Sena A L S, Santos M A S dos, Santos J C e Homma A K O 2012 Avaliação do nível tecnológico dos produtores de leite na região oeste do estado do Pará. Revista de Economia e Agronegócio, 10 (3): 397-418. 

Solano C, Bernués A, Rojas F, Joaquín N, Fernández W and Herrero M 2000 Relationship between management intensity and structural and social variables in dairy and dual-purpose systems in Santa Cruz, Bolivia. Agricultural. Systems, 65 (3): 159–177. 

Tsioulpas A, Grandison A S and Lewis M J 2007 Changes in physical properties of bovine milk from the colostrum period to early lactation. J. Dairy Science, 90 (11): 5012–5017. 

Vargas-Leitón B, Guzmán O S, Segura F S y Hidalgo H L 2013 Caracterización y clasificación de hatos lecheros en costa rica mediante análisis multivariado. Agronomía Mesoamericana, 2 (2): 257-275. Retrieved december 30, 2013, from: http://revistas.ucr.ac.cr/index.php/agromeso/article/view/12525/11767

Received 28 January 2016; Accepted 8 May 2016; Published 2 June 2016

Go to top