Keywords:

Theobroma cacao L.

Dimensional indexes

Clustering

Logistic regression

Functional relationships and productivity factors of the ne aroma cocoa production system

Relaciones funcionales y factores de productividad del sistema de producción de cacao no de

aroma

Relações funcionais e fatores de produtividade do sistema de produção de cacau de aroma no

Julia Martínez Sthormes

Fátima Urdaneta Ortega

María Elena Peña

Ángel Casanova Araque

1†

Rev. Fac. Agron. (LUZ). 2026, 43(2): e264320

ISSN 2477-9407

DOI: https://doi.org/10.47280/RevFacAgron(LUZ).v43.n2.II

Socioeconomics

Associate editor: Dra. Maritzabel Materán Jaimes

University of Zulia, Faculty of Agronomy

Bolivarian Republic of Venezuela.

Universidad del Zulia. Facultad de Agronomía. División de

Estudios para Graduados, Maracaibo, Venezuela.

Universidad del Zulia. Facultad de Ciencias Veterinaria.

División de Estudios para Graduados.

Received: 23-12-2025

Accepted: 16-03-2026

Published: 31-03-2025

Abstract

In the South of Lake Maracaibo, ne aroma cocoa production

systems (FACPS) have been developed where good quality

beans are produced, but with low yields; this situation can be

associated with multiple social, technical and economic factors,

whose interrelationships require a comprehensive analysis to be

able to identify their limitations and potentialities. This research

was proposed with the objective of explaining how the functional

relationships of the FACPS aect its productivity, as well as

weighing the productive factors. A sample of 84 producers from

the South of Lake Maracaibo was taken. Data were analyzed by

Cluster analysis to group the production units by their functional

similarities according to four calculated indices: Agronomic

Practices Index (API), Labor Force Index (LFI), Production Means

Index (PMI) and Socioeconomic Environment Index (SEI), then

an analysis of variance was performed to establish the productivity

dierences between groups. The factors weighting was carried out

by a logistic regression model. Results showed the formation of

four functional groups with their arrangement of components and

relationships. Productivity indicators were signicantly dierent

(p ≤ 0.05) among groups, this indicates that the way the system’s

components are arranged aects its productive outputs. The logistic

regression model indicated that the educational level, the low

percentage of at surface and the farm size are the main factors that

increase the probability that a production unit belongs to the group

with the highest yields.

This scientic publication in digital format is a continuation of the Printed Review: Legal Deposit pp 196802ZU42, ISSN 0378-7818.

Rev. Fac. Agron. (LUZ). 2026, 43(2): e264320 April-June ISSN 2477-9409.

2-6 |

Resumen

En el Sur del Lago de Maracaibo, se han desarrollado sistemas de

producción de cacao no de aroma (SPCFA) los cuales producen granos

de buena calidad, pero con rendimientos bajos, esta situación puede

asociarse a múltiples factores sociales, técnicos y económicos, cuyas

interrelaciones requieren de un análisis integral para poder identicar

sus limitaciones y potencialidades. Se planteó esta investigación

con el objetivo de explicar cómo las relaciones funcionales del

SPCFA inciden en su productividad, asimismo ponderar los factores

productivos. Para ello se tomó una muestra de 84 productores del

sur del lago de Maracaibo, los datos se analizaron por medio de un

análisis Clúster para agrupar las unidades de producción por sus

similitudes funcionales de acuerdo a cuatro índices calculados: Índice

de Prácticas Agronómicas (IPA), Índice de fuerza de trabajo (IFT),

Índice de Medios de Producción (IMP) e Índice de Entorno (IEN),

luego se realizó análisis de varianza para establecer las diferencias

de la productividad entre grupos. La ponderación de factores se

realizó por medio de un modelo de regresión logística. Los resultados

muestran la identicación de cuatro grupos de arreglo de componentes

y relaciones, los indicadores de productividad resultaron diferentes

signicativamente (p ≤ 0,05) para los grupos funcionales, lo que

indica que la forma como se relacionan los componentes del sistema

incide en sus salidas productivas. El modelo de regresión logística

indicó que el nivel educativo, el bajo porcentaje de supercie plana y

el tamaño del predio, son los principales factores que incrementan la

probabilidad de que una unidad de producción pertenezca al grupo de

mayores rendimientos.

Palabras clave: Theobroma cacao L., índices dimensionales, clúster,

regresión logística.

Resumo

No Sul do Lago de Maracaibo, se foram desenvolvidos sistemas

de produção de cacau no de aroma (SPCFA), mas com desempenhos

baixos, os quais podem ser associados a vários fatores sociais,

técnicos e econômicos, cujas inter-relações exigem uma análise

integral para poder identicar suas limitações e potencialidades. Se

você planejasse esta investigação com o objetivo de explicar como

as relações funcionais do SPCFA incidem em sua produtividade,

o simismo pondera os fatores produtivos. Para que ele tenha uma

mostra de 84 produtores do sul do lago de Maracaibo, os dados são

analisados por meio de uma análise. Clúster para agrupar as unidades

de produção de acordo com suas semelhanças funcionais de acordo

com quatro índices calculados: Índice de Práticas Agronômicas (IPA),

Índice de Força de Trabalho (IFT), Índice de Meios de Produção

(IMP) e Índice de Entorno (IEN), depois foram realizadas análises de

variação para estabelecer as diferenças de produtividade entre grupos.

A ponderação de fatores foi realizada por meio de um modelo de

regressão logística. Os resultados mostram a identicação de quatro

grupos de acúmulo de componentes e relações, os indicadores de

produtividade resultam signicativamente diferentes (p ≤ 0,05) para

os grupos funcionais, o que indica que a forma como se relaciona

com os componentes do sistema incide em seu saídas produtivas. O

modelo de regressão logística, que indica que o nível educativo, a

baixa porcentagem da superfície plana e o tamanho do preço, são os

principais fatores que aumentam a probabilidade de que uma unidade

de produção pertença ao grupo de maiores rendimentos.

Palavras chave: Theobroma cacao L., índices dimensionais,

agrupamento, regressão logística.

Introduction

In many tropical countries, cacao is cultivated by small-scale

producers who rely primarily on family labor (Arvelo et al., 2017),

and the southern region of Lake Maracaibo is not an exception. It is

one of the most important agricultural commodities in international

trade and, as such, an indispensable source of foreign exchange for

producing countries. Furthermore, it is one of the most dynamic

sectors in rural areas, attracting the interest of international markets.

The importance in Venezuela lies in its contribution of 8.5 % to the

international ne cacao market; however, it only accounts for 1 % of

total world production (Fundacite-Zulia, 2007).

Historically, cocoa cultivation in the southern Lake Maracaibo

region (SLM) has been traditionally developed in peasant production

systems, which have contributed to increasing the income level of

rural families without compromising the stability of ecosystems.

Furthermore, excellent quality cacao is produced (Portillo-López et

al., 2025), although this production has not been able to surpass the

world average yield of 385 kg.ha

-1

(FAO, 2020).

These production systems are considered a sustainable alternative

to intensive agricultural systems characterized by monoculture

(Espinosa-Álzate & Ríos-Osorio, 2016). Structural elements of the

production systems are closely related, so situations arising in one

component aect the development of the entire system.

The low production of the SLM ne aroma cocoa production

system can be associated with its low technological level, as well

as the lack of training, scarce labor, high genetic variability in its

plantations, lack of economic incentives and inecient infrastructure

(Quintero, 2020).

It is also reported that the cocoa producer carries out very few

cultural practices. such as fertilization, pruning, weed control, and pest

and disease management, negatively impacting quality, yield, and,

consequently, their competitiveness in dierent markets (Zambrano-

Piña and Segovia-López, 2011). Similar to other Latin American

countries, farms exhibit deciencies in post-harvest management,

where fermentation is carried out in a very low proportion. (Guillén

et al., 2023).

It is evident that there are many factors involved in the production

process of ne aroma cocoa, which in turn aect its supply, making

it essential to understand the aspects that characterize the production

system of cocoa-growing areas.

Given the complex combination of variables and interrelationships

that occur in cocoa production systems, it is necessary to identify

the factors that inuence their production and productivity from

a comprehensive perspective, considering the methodological

basis of the systems approach (Casanova et al., 2016). This will

allow researchers to design improvement strategies to address any

diculties that may arise in the production unit.

Considering the above, the hypothesis was put forward that the

existing variations in the productivity of ne aroma cocoa production

systems (SPCFA) respond to the functional relationships that occur

between critical factors, such as agronomic practices, means of

production, labor force and socioeconomic environment factors.

This scientic publication in digital format is a continuation of the Printed Review: Legal Deposit pp 196802ZU42, ISSN 0378-7818.

Martínez et al. Rev. Fac. Agron. (LUZ). 2026, 43(2): e264320

3-6 |

According to this approach, the objectives were formulated to

explain how the functional relationships of the ne aroma cocoa

production system in the south of Lake Maracaibo determine its

production and productivity and to weigh the factors that aect the

production and productivity of said system.

Materials y methods

This research work was oriented under the epistemic empirical

inductive approach (Padrón, 2007) reaching an explanatory level

which determines its type and with an ex post facto, cross-sectional

design supported by eld data, since the researcher does not manipulate

or control variables, the information is taken at a single cut in time

and is obtained from a natural environment (Hurtado, 2010).

The population considered for this study was the result of a census

conducted prior to sample collection, which included 452 cocoa

producers located in the southern part of Lake Maracaibo, in the

municipalities of Tulio Febres Cordero, Julio Cesar Salas, and Justo

Briceño in the state of Mérida, as well as the municipality of Sucre

in the state of Zulia. These municipalities are located along the Pan-

American Highway between 71° 30’ 20” and 71° 45’ 5” W longitude

and 8° 20’ 36” and 8° 40’ 12” N latitude.

The area features natural vegetation characteristic of tropical

rainforest and premontane forest. Rainfall ranges from 1,000 mm in

the north to 1,800 mm in the south. The altitude ranges from 40 to 100

meters above sea level in the atlands and from 100 to 1,000 meters

above sea level in the foothills and mountains. The average monthly

temperature ranges from 20 to 28 °C. The soils in the area are deep

and well-drained (Gómez & Azócar, 2002).

The sample was selected using stratied random sampling with

proportional allocation, where strata were structured based on the

municipality and size of the cacao production unit. The sample size

was 90 production units, with a 95 % condence level and a 7 %

sampling error, but due to data inconsistencies, the nal sample

consisted of 84 production units (Table 1). The SAS surveyselet

procedure and the Proportional Selection Probabilities (PPS) method

for stratied samples without replacement were used (Tamayo, 2000).

Tabla 1. Distribución de la muestra por estratos.

Municipality

Estrata

Tulio

Febres

(n)

Julio

César

Salas

(n)

Justo

Briceño

(n)

Sucre

(n)

Total

selected

Denitive

Sample

< 1 ha 28 14 0 2 44 40

1 - 5,5 ha 18 16 2 0 36 34

5,5 – 10,5 ha 3 1 1 0 5 5

10,5 – 15,5 ha 2 1 0 0 3 3

>15,5 ha 1 1 0 0 2 2

Total 52 33 3 2 90 84

N= number of production units (PU).

Data collection techniques

Field information was collected through a self-developed

questionnaire, based on the systematization of the study variable

(Table 2), according to the approaches of: Dufumier (2014); FAO &

BM (2002); Morín (2004); Osorio (2012); Quintero & García (2010);

Soler, (2017); Uzcátegui (2020).

Information was also collected on indicators for calculating

productivity: production (kg), harvested area (ha), productivity (kg.

-1

), and estimated income (Bs). The validity of the questionnaire

was assessed through expert judgment, and its reliability was assessed

using the test-retest reliability method (Berchtold, 2016), which

yielded a reliability coecient of 0.90.

Table 2. Systematization of the system functional relationships

variable.

Variable Dimension Indicator

Functional

relationships of

the ne aroma

cocoa production

system (FACPS)

Agronomic

practices

Type of cacao planted. Age of

the plantation (years). Number

of plants. Area planted. Planting

density. Fertilization. Weed control.

Pest control. Irrigation. Pruning.

Drying. Fermentation.

Labor Force

Producer’s age. Educational level.

Type and number of workers:

Family-employed or hired.

Permanent or temporary. Sta

training. Technical assistance.

Means of

production

Land distribution and use.

Topography. Texture. Water source

and use. Type, age, and condition of

buildings and facilities. Machinery

and equipment.

Socioeconomic

environment

Agricultural Support: Credit,

availability, terms. Agricultural

Services: Availability of technical

services. Agricultural Businesses:

Availability of inputs, buying

and selling companies, collection

centers, trade channels.

Public services. Land tenure

security.

Data Analysis Techniques

Dimensional analysis is a powerful analytical tool designed to

nd or verify relationships between physical quantities by using

their dimensions

(Moeenifar et al., 2013). That is, the construction of

several indices from the indicators of each study variable dimension.

The indices for Agronomic Practices (API), Labor Force (LFI),

Means of Production (MPI), and socioeconomic environment (SEI)

have dierent natures due to the units in which their indicators are

expressed.

Indicators whose nature yielded a “yes” or “no” response were

measured with values of one (1) for the presence of the attribute or

zero (0) for its absence. When the value was continuous quantitative,

all records were divided by the highest value to bring them all to

a scale between 0 and 1, which allowed the calculation of each

dimensional index as the mean of the total value of the indicators that

make up each one.

Using these values, a multivariate analysis (K-Means Cluster)

was performed, which allowed the grouping of the most similar

production units and the selection of the groups with the greatest

distance between them in the multidimensional space (Wackerly et

al., 2009). Then, an analysis of variance was performed to determine

the dierence in production and productivity between the groups.

Finally, a weighting analysis of production factors was performed

using a logistic regression model. To improve the model’s accuracy,

an outlier analysis was conducted to remove extreme values, resulting

in 77 UP for the model. Logistic regression is a model that, based

on the estimated coecients of a set of independent variables and

associating a probability that the dependent variable takes a value of

1, allows each individual to be assigned to one category or another.

As with the discriminant function, the model will have a good t and

This scientic publication in digital format is a continuation of the Printed Review: Legal Deposit pp 196802ZU42, ISSN 0378-7818.

Rev. Fac. Agron. (LUZ). 2026, 43(2): e264320 April-June ISSN 2477-9409.

4-6 |

be predictive if it correctly classies individuals into their respective

categories with a high degree of accuracy (Alderete, 2006).

The mathematical expression that represents the model is the

following:

Let P (Y = 1/X) be the probability that y = 1, given the vector X

of independent variables, the form of the logistic regression model is:

P (Y= 1/X) = 1/(1+ e –z)

Where:

Z= Bo +B1X1+B2X2+… +BpXp

Bo, B1, B2… Bp are lineal regression coecients

X1, X2… Xp are independent variables

e = base of natural logarithms (2,718)

The productivity variable expressed in kg.ha

-1

(Yield), was

selected to divide the sample into two groups: Group 1 made up of

the farms that achieved a yield below the median, with a coded value

Y= 0 (low yields) and Group 2 made up of all the farms that achieved

yields above the median with a value Y= 1 (high yields).

The independent variables in the logistic regression model

were considered in sets (blocks), each corresponding to one of the

dimensions of the production system addressed in this study. To ensure

their inclusion in the model, the backward stepwise variable selection

method was used, with entry and exit signicance levels set at 0.20 and

0.30, respectively. These levels are considered conservative for non-

experimental eld data, where error values are expected to be higher

than those expected with experimental data. Data matrix was structured

in Excel, and analyses were performed using SPSS version 21.

Results and discussion

Grouping farms is useful because future improvement actions

can be designed and implemented by groups rather than individually

which saves resources. In order to explain how the functional

relationships within the ne aroma cocoa production system aect

production and productivity, It was necessary rst to study how the

structural elements interrelate to create the product.

Arrangement of functional groups

Four functional groups were obtained (Table 3), as this number of

groups demonstrated robust classication, optimizing the distribution

of systems within them to minimize the sum of the distances of each

observation from the center of its group. These groups are relatively

homogeneous within themselves and heterogeneous with respect to

each other based on a dened set of variables (Guillén et al., 2023).

The rst functional group (G1) consisted of 25 % of the production

units (n=21), the second group (G2) was formed by the 22.62 %

(n=19), G3 by 21.43 % (n=18) and nally, the largest group (G4)

collected the 30.95 % of the farms (n=26).

Table 3. Final values for each cluster center (Functional Group).

Indices

Cluster (group)

(n=21)

(n=19)

(n=18)

(n=26)

API**

0.42

0.65

0.40 0.38

LFI*

0.44 0.45

0.46

0.41

PMI**

0.67

0.71

0.48 0.69

SEI*

0.29

0.61

0.56

0.61

API: Agronomic Practices Index. LFI: Labor Force Index. PMI: Production Means Index.

SEI: Socioeconomic Environment Index. *Signicant dierence P≤0,05; **Signicant

dierence P≤0,01.

As can be seen in Table 3, there were signicant dierences

(P≤0.05 and P≤0.01) between the groups for all calculated indices.

The F-test is performed for descriptive purposes since the clusters

were intentionally chosen to maximize the dierences between cases

in dierent clusters.

Group G1 is characterized by having the lowest Environmental

Index (SEI=0.29) and average values for the Agronomic Practices

Index (API=0.42), Workforce Index (LFI=0.44), and Means of

Production Index (PMI=0.67). This group exhibits weaknesses

in its relationship with the environment and average agronomic

performance, typical of the area. These results align with those

published by Mata et al. (2018), where average and low index values

indicate limitations in their relationships whit the socioeconomic

environment and production process, as well as dependence on family

labor and limited technical training.

Group G2 stands out for showing the highest values for API

(0.65), PMI (0.71), and SEI (0.61), making it the group with the

highest values in most indices; LFI (0.45) achieves the second-highest

value across all groups. This group’s functional relationships are the

best, as higher values in these indices indicate improved execution of

both their agronomic plan and the adequate availability of production

resources, as well as their interactions with the socioeconomic

environment.

Group G3 is characterized by having the best LFI value (0.46),

but conversely, the lowest PMI value (0.48), indicating deciencies

in this aspect. The higher LFI value indicates better sta training,

producer’s educational level, and technical assistance, among other

factors. These results are comparable to those of producers in the

central micro region of Manabí, Ecuador, who demonstrated potential

in their workforce by showing the highest frequency of primary

education, dedication to cultivation, overnight stays at the production

unit, and access to technical assistance (Guillén et al. 2025).

Finally, G4, whose strength lies in the SEI (0.61), sharing the lead

with G2, is noteworthy for its lower API (0.38) and LFI (0.41) values.

This group exhibits weaknesses in its production process and human

resources, characterized by low levels of education and limited

training, but with strong commercial relationships (SEI=0,60). These

characteristics are similar to those reported by Mata et al. (2018), who

also conrm the deciencies in the means of production available to

cocoa farmers.

Production and productivity by functional group

Table 4 shows means and standard deviations of the production

and productivity indicators by functional group, where the analysis of

variance reveals signicant dierences (P ≤ 0.05) for all the indicators

studied: total cocoa production (TCP), yield per area (YHA), total

production area of the farm (TFA), cocoa production area (CPA) and

total income (TIN).

The best yields were observed in G2 (group with the highest

API, PMI, and SEI) and G3 (group with the highest LFI), such

that the dierences between the functional groupings also show

statistically signicant dierences for both total production and

productivity (YHA). These results are similar to those found by

Guillen et al. (2023) regarding statistically signicant dierences

(p ≤ 0.05) between the functional producer groups, both in total

cocoa production and in yield per hectare and other economic

variables, thus conrming the importance of combining functional

factors appropriate to the production and quality of the crop, such as

agronomic and post-harvest management. The average cocoa yields

and economic indicators allow us to identify regions with potential to

This scientic publication in digital format is a continuation of the Printed Review: Legal Deposit pp 196802ZU42, ISSN 0378-7818.

Martínez et al. Rev. Fac. Agron. (LUZ). 2026, 43(2): e264320

5-6 |

increase the area, production, and current competitiveness of this crop

(Espinosa, et al., 2015).

Table 4. Average values of production and productivity indicators

for each functional group.

GROUP

TCP**

(kg)

YHA* (kg.

-1

)

CPA*

(ha)

TFA*

(ha)

TIN*

(Bs)

Mean 940.24 329.65 2.73 3.03 1.455.76

SD 1.162.59 118.78 2.41 2.84 1.953.27

Mean

2.640.86 470.22 5.10 5.63 5.088.58

SD 3.992.36 318.25 4.22 4.92 9.184.56

Mean 822.61 408.70 2.18 2.32 1.274.88

SD 4163.60 318.25 2.14 2.14 1090.63

Mean 814.96 330.28 2.75 2.90 1.234.78

SD 843.57 140.23 3.27 3.21 1.322.43

Total

Mean 1260.92 378.58 3.16 3.43 2.170.31

SD 2159.81 200.02 3.26 3.57 4.374.53

*Signicant dierence P≤0,05; **Signicant dierence P≤0,01. SD: Standar

deviation. TCP: Total cocoa production. YHA: Yield per hectare. CPA: Cocoa

production area. TFA: Total farm area TIN: Total Income.

The high standard deviations in income are noteworthy, especially

in group 2. This indicates very high coecients of variation, likely

explained by the price dierences these producers receive, which

aect both the mean and the total standard deviation. These producers

have the best relationships with their environment (including the

market and marketing), the highest yields, total production, planted

area, and probably the best quality cocoa produced, as evidenced by

the higher value of the Agronomic Practices Index (API), which also

includes post-harvest management.

Factors weighting

From the rst set of variables analyzed by the logistic regression

model, the following were selected to be included in the equation:

Cacao Area (CA), Percentage of Flat Area (FAP), and Presence of

Forastero “Pajarito” Type Cocoa (FPTP). From the second set, the

variables included in the equation were: Educational Level, Family

Labor Personnel, and Housing Index (EDUL, FAML, and HOUSI).

None of the variables from the third set were included in the equation,

then agronomic management variables were eliminated. Finally, only

Credit variable (CRED) was included from the fourth set.

Table 5 shows that 23 UPs from the low-yield group were correctly

predicted, which represents 65.71 % of that group. In the high-yield

group, 31 UPs were correctly predicted, representing 73.8 % of the

group. Based on the total sample (n=77), 54 UPs were correctly

categorized, representing 70.1 % of the farms. Consequently, it can be

stated that the model t is acceptable, given that data were collected

without manipulation or control of variables. A good discrimination

threshold is typically 50 % (corresponding to odds=1), meaning that

the two response options are equally likely. By setting this threshold,

it is considered that if the response is positive and the estimated

probability is greater than 50 %, then the model is enough accurate

(Guillen & Alonso, 2020).

Table 6 shows results of the logistic regression analysis. The rows

list the variables included in the model, and the columns show the

regression coecient (B) associated with each variable, the standard

error (SE), the Wald test criterion degrees of freedom (df), the

signicance level (Sig.), and the odds ratio (Expb).

The most important factor in the model, with a signicance value

less than 5 % , was the producer’s educational level (NEDUC), with

a positive B value.

Table 5. Frequency values, observed and predicted by the model,

for each yield group. (n= number of production units).

Predicted (n)

Group

Low

Yield

High

yield

Total

(n)

Percentage of farms

correctly classied

(%)

Observed

(n)

Low Yield 23 12 35 65.71

High Yield 11 31 42 73.80

Total 34 43 77 70.12

The odds ratio is interpreted as the number of times the probability

of nding an individual in the y=1 category (high yield group)

increases or decreases compared to the y=0 category (Low yield

group) when the variable increases by one unit. Mathematically, this

is expressed as P(Y=1/x)/P(y=0/x).

Table 6. Variables in the Logistic Regression Equation.

Variables B S.E. df Sig. Exp (B)

EDUL

1.680 0.642 1 0.009** 5.367

FAP

-0.016 0.009 1 0.064* 0.984

COCOA

-0.141 0.084 1 0.094* 0.869

FAML

0.481 0.313 1 0.125 1.618

PTFP

0.780 0.602 1 0.195 2.182

HOUSI

3.898 2.694 1 0.148 49.288

CRED

-1.197 0.883 1 0.175 0.302

CONSTANT

-7.017 3.463 1 0.012 0.000

EDUL: Educational level, FAP: Flat surface percentaje, COCOA: Cocoa area FAML: Family

labor, PTFP: “Pajarito” type foreign cacao presence, HOUSI: House index CRED: Credit.

This can be interpreted as meaning that as the producer’s

educational level improves, the probability of the farm belonging

to the group of highest-yielding producers increases. In the Exp (B)

column, the corresponding value is 5.367, which means that for each

unit that educational level increases, the likelihood of the farm to

belong to the high yield group increases more than vefold. These

results are corroborated by Kabiru (2020), who reports a positive

relationship between education and agricultural productivity.

The second most important factor (Table 6) is the variable

PFA, with a signicance level of 0.064 and a negative B value,

indicating that an increase in at land area decreases the probability

of nding farms with better yields. Farms located in areas with more

undulating landscapes (with at land areas less than 30 %) oer better

temperature, soil, and rainfall conditions for cocoa cultivation than

lower, atter areas.

A third important factor in the model was the area planted with

cacao (COCOA), which showed a signicance level of 0.094 with

a negative B value, indicating that as the area planted with cacao

increases, the probability of nding a high-yielding farm decreases.

The average planted area per farm is 3.5 ± 2.84 ha, although many

small farms of 1 ha, and even as small as 0.50 ha, exist. These farms

do not require high maintenance and management costs, as family

labor is sucient to perform the necessary tasks. Therefore, it is

This scientic publication in digital format is a continuation of the Printed Review: Legal Deposit pp 196802ZU42, ISSN 0378-7818.

Rev. Fac. Agron. (LUZ). 2026, 43(2): e264320 April-June ISSN 2477-9409.

6-6 |

expected that these small plantations are better managed, which will

be reected in higher yields. Similarly, among the factors that explain

yield in most of the cantons of the Guayas Province of Ecuador are

the number of hectares planted and the labor force (Quinde et al.,

2019), which is primarily family-based.

In this regard, it was found that the next most important factor

is family labor (FAML), with a signicance level of 0.125 and a

positive B value, indicating that a higher proportion of family labor

increases the probability of nding high-yield farms. The probability,

as indicated by Exp (B), is 1.618 per unit of family labor incorporated

into the system. The last three factors included in the model PFTP,

HOUSI, and CRED had signicance levels ranging from 0.15 to 0.20,

which are still acceptable for non-experimental data. The presence of

the “pajarito” cacao type (PTFP) the housing index (HOUSI) both

have positive B values, suggesting a higher probability of nding

high-yield farms with a greater presence of these two factors.

Conclusions

Results of the functional analysis showed four distinct component

arrangement groups with dierent relationships between them.

Furthermore, all production and productivity indicators diered

signicantly (p≤0.05) between the functional groups, indicating that

the way the system’s components are related aects the system’s

productive outputs, thus conrming the hypothesis that guided this

research.

Prole of G2 stands out, showing the best values in most of the

studied indices API (0.65), PMI (0.71) and SEI (0.61); as well as the

highest values in all productive response indices, demonstrating that

the application of agronomic practices, the availability of adequate

means of production and the relationships with the environment, are

fundamental for the improvement of production, productivity and

income.

According to the logistic regression model results, a ne aroma

cocoa production system will have a high probability of belonging

to the high-yield group if, rstly, it is managed by a highly educated

producer; secondly, it has less than 5 hectares of planted cocoa and a

at area percentage less than 30 %; thirdly, it has family workers, a

high housing index, and has introduced “pajarito” type foreign cocoa;

and nally, it does not need access to agricultural credit.

However, in light of these results, the implementation of an

eective plan for ne aroma cacao must be a priority. This plan aims

to rescue and propagate these varieties through awareness-raising

and training cacao farmers on the importance of good agricultural

and post-harvest practices. Such practices will allow for higher levels

of productivity and quality, preventing their replacement by foreign

cocoa types like “Pajarito”, which are more rustic and less aromatic,

but oers higher yields. A policy is also needed to value “Criollo”

cacao, whose quality commands a higher price due to its greater

demand in international markets.

Literature cited

Alderete, A. (2006). Fundamentos del Análisis de Regresión Logística en la

Investigación Psicológica. Evaluar, 6, 52–67. https://doi.org/10.35670/1667-

4545.v6.n1.534

Arvelo, M., Gonzáles, D., Delgado, T., Maroto, S., & Montoya, P. (2017). Estado

actual sobre la producción, el comercio y cultivo del cacao en América.

IICA-Fundación Colegio para postgraduados. https://blog.fundacioncolpos.

org/?projects=estado-actual-sobre-la-produccion-el-comercio-y-cultivo-del-

cacao-en-america

Berchtold, A. (2016). Test–retest: Agreement or reliability? Methodological

Innovations, 9, 1–7. https://doi.org/10.1177/2059799116672875

Casanova, L., Martínez, J., López, S., & López, G. (2016). De von Bertalany

a Luhmann: Deconstrucción del concepto "agroecosistema" a través

de las generaciones sistémicas. Revista MAD, 35, 60–74. https://doi.

org/10.5354/0718-0527.2016.42797

Dufumier, M. (2014). Agriculturas familiares, fertilidad de los suelos y sostenibilidad de

los agroecosistemas. In P. Hidalgo, F. Houtart, F. Lizárraga (Ed.), Agriculturas

campesinas en Latinoamérica: propuestas y desafíos. Editorial IAEN.

https://www.ippri.unesp.br/Modulos/Noticias/328/agriculturas-campesinas-

propuestas-y-desaos-web.pdf

Espinosa, J., Urestil, J., Vélez, A., Moctezuma, G., Inurreta, H., y Góngora, S. (2015).

Productividad y rentabilidad potencial del cacao (Theobroma cacao L.) en el

trópico mexicano. Revista mexicana de ciencias agrícolas, 6(5), 1051-1063.

https://cienciasagricolas.inifap.gob.mx/index.php/agricolas/article/view/

598/471

Espinosa-Álzate, J. A., & Ríos-Osorio, L. A. (2016). Caracterización de sistemas

agroecológicos para el establecimiento de cacao (Theobroma cacao L.), en

comunidades afrodescendientes del Pacíco Colombiano (Tumaco- Nariño,

Colombia). Acta Agronómica, 65(3), 211–217. https://doi.org/10.15446/acag.

v65n3.50714

FAO, & BM. (2002). Sistemas de producción agropecuarios y pobreza. FAO and

World Bank. https://www.fao.org/3/y1860s/y1860s03.htm#TopOfPage

FAO (2020). FAOSTAT Database Results. http://www.fao.org.

Fundación para la Ciencia y la Tecnología del Estado Zulia (Fundacite-Zulia). (2007).

Hacia el Plan Nacional del Cacao. Memorias: I Encuentro Nacional Ruta Del

Chocolate.

Gómez, A., y Azócar, A. (2002) Áreas potenciales para el desarrollo del cultivo cacao

en el Estado Mérida. Agronomía Tropical, 52(4), 403-425. http://ve.scielo.

org/scielo.php?script=sci_arttext&pid=S0002-192X2002000400001&lng=e

s&nrm=iso

Guillén, M., & Alonso, M. (2020). Modelos de regresión logística (2da ed.). Universitat

oberta de Catalunya. https://openaccess.uoc.edu/server/api/core/bitstreams/

e830ec11-40f0-4aec-bc45-8bc10b123f7e/content

Guillén, S., Urdaneta, F., & Portillo, E. (2023). Tipicación socioproductiva

de los agroecosistemas de producción de cacao de Manabí, Ecuador.

Agroalimentaria, 29(56), 65–81. http://erevistas.saber.ula.ve/index.php/

agroalimentaria/article/view/19290/2023.01.56.04

Guillén, S., Urdaneta, F., & Portillo, E. (2025). Caracterización socio-técnica-

productiva de los sistemas de producción de cacao en las microrregiones de

la provincia de Manabí, Ecuador. Revista de la Facultad de Agronomía de la

Universidad de la Zulia, 42(4), 2-7. https://www.produccioncienticaluz.org/

index.php/agronomia/article/view/44792

Hurtado, J. (2010). Metodología de la Investigación. Guía para la comprensión

holística de la Ciencia (4ta ed.). Ediciones Quirón.

Kabiru, S. (2020). The Relationship Between Education and Agricultural Productivity :

The Moderating Eect of NGO. Universal Journal of Educational Research,

8(3), 866–871. https://doi.org/10.13189/ujer.2020.080317

Mata, D; Rivero, M., y Segovia, E. (2018) Sistemas agroforestales con cultivo

de cacao no de aroma: entorno socioeconómico y productivo. Revista

Cubana Ciencias Forestales, 6 (1), 103-115. http://scielo.sld.cu/scielo.

php?script=sci_arttext&pid=S2310-34692018000100103&lng=es&nrm=i

so>. ISSN 2310-3469.

Moee

nifar, A. M., Kalantari, D., Reza, S., & Mousavi, S. (2013). Application

of dimensional analysis in determination of traction force acting on a

narrowblade. International Journal of Agriculture and Crop Sciences, 5(9),

1034–1039. www.ijagcs.com.

Morín, E. (2004). Introducción al pensamiento complejo. GEDISA. https://

cursoenlineasincostoedgarmorin.org/images/descargables/Morin_

Introduccion_al_pensamiento_complejo.pdf

Osorio, S. (2012). El pensamiento complejo y la transdisciplinariedad: fenómenos

emergentes de una nueva racionalidad. Revista Facultad de Ciencias

Económicas: Investigación y Reexión, 20 (1), 269–291. www.scielo.org.co/

pdf/rfce/v20n1/v20n1a16.pdf

Padrón, J. (2007). Tendencias Epistemológicas de la Investigación Cientíca en

el Siglo XXI. Cinta de Moebio: Revista Electrónica de Epistemología de

Ciencias Sociales, 28, 1–28. www.moebio.uchile.cl/28/padron.html

Portillo-López, A., Portillo, E., Cohen, J., & Urdaneta, F. (2025). Sistema

georreferenciado para la trazabilidad de la calidad del cacao ( Theobroma

cacao L .). In R. Meleán & Y. Montes de Oca (Eds.), Agronegocios

Inteligentes (pp. 87–96). Universidad del Zulia/High Rate Consulting. https://

doi.org/https://doi.org/10.38202/agronegocios

Quintero, M. (2020). La cadena agroalimentaria del cacao en Venezuela : hacia el

desarrollo territorial. Agroalimentaria, 26, 213–237. https://dialnet.unirioja.

es/servlet/articulo?codigo=7998166

Quintero, M., & García, L. (2010). La producción de cacao en Venezuela. Hacia una

nueva ruralidad. Actualidad Contable FACES, 13(20), 114–123. http://www.

redalyc.org/articulo.oa?id=25715828009

Soler, Y. (2017). Teorías sobre los sistemas complejos. Revista Administración &

Desarrollo, 47(2), 52–69. https://doi.org/10.22431/25005227.201

Tamayo, G. (2000). Diseños muestrales en la investigación. Economic Semester, 4(7).

https://revistas.udem.edu.co/index.php/economico/article/view/1410/1542

Uzcátegui, J. (2020). Pensamiento sistémico y teoría de la complejidad en el análisis

de sistemas agropecuarios sostenibles: conciencia metódica. FERMENTUM.

Revista Venezolana de Sociología y Antropología, 30(88), 237–283. https://

hdl.handle.net/20.500.14330/CLA01000499878

Wackerly, D., Mendenhall III, W., & Scheaer, R. (2009). Estadística matemática con

aplicaciones (S. Cervantes (ed.); 7ma ed.). Cengage learning. http://www.

economia.unam.mx/biblioteca/Pdf/Estadística matemática con aplicaciones

7edi_MENDENHALL.pdf

Zambrano-Piña F., y Segovia-López E. (2011). La competitividad del sistema de

producción de cacao en la zona norte del estado Táchira. Revista de la

Facultad de Agronomía de la Universidad de la Zulia, 28(4), 566-595.

https://produccioncientifica.luz.edu.ve/index.php/agronomia/article/

view/26898/27524