sustainability

Article
Understanding Farm Diversity to Promote
Agroecological Transitions
Heitor Mancini Teixeira 1,2,3, * , Leonardo van den Berg 4,5 , Irene Maria Cardoso 2 ,
Ardjan J. Vermue 1 , Felix J. J. A. Bianchi 1 , Marielos Peña-Claros 3 and Pablo Tittonell 6,7
1 Farming Systems Ecology Group, Wageningen University & Research, P.O. Box 430, 6700 AK Wageningen,
The Netherlands; ardjan.vermue@gmail.com (A.J.V.); felix.bianchi@wur.nl (F.J.J.A.B.)
2 Department of Soil and Plant Nutrition, Federal University of Viçosa, campus UFV, 36570-900 Viçosa, Brazil;
irene@ufv.br
3 Forest Ecology and Forest Management Group, Wageningen University & Research, P.O. Box 47,
6700 AA Wageningen, The Netherlands; marielos.penaclaros@wur.nl
4 Forest and Nature Conservation Policy, Wageningen University & Research, P.O. Box 47,
6700 AA Wageningen, The Netherlands; leonardo.vandenberg@wur.nl
5 Department of Rural Economics, Federal University of Viçosa, campus UFV, 36570-900 Viçosa, Brazil
6 Agroecology, Environment and Systems Group, Instituto de Investigaciones Forestalesy Agropecuarias de
Bariloche (IFAB), INTA-CONICET, Modesta Victoria 4450-CC 277, 8400 San Carlos de Bariloche, Río Negro,
Argentina; tittonell.pablo@inta.gob.ar
7 Groningen Institute of Evolutionary Life Sciences, Groningen University, P.O. Box 11103,
9700 CC Groningen, The Netherlands
* Correspondence: heitor.manciniteixeira@wur.nl; Tel.: +55-31-997-138-377




Received: 12 September 2018; Accepted: 17 November 2018; Published: 22 November 2018


Abstract: Agroecology is increasingly promoted by scientists, non-governmental organisations

(NGO’s), international organisations and peasant movements as an approach to foster the transition
to sustainable and equitable food systems. The challenges to agroecological transitions are
not the same for all farmers, as they can face different social and bio-physical conditions.
We developed a farm typology combining participatory and quantitative methodologies to assess
and categorise farm diversity and its implications for developing strategies to promote agroecological
transitions. The participatory typology was developed during workshops to acquire insights on
local farmers’ perceptions and knowledge, and to generate hypotheses on family farm diversity.
The participatory-based hypotheses were tested in the quantitative farm characterisation, which
provided information on household characteristics, production strategies, land use, participation
in public policies and extension services. Farms were located in Zona da Mata, Minas Gerais,
Brazil, which harbour a wide diversity of farmers and where different actors have been engaged
in agroecological transitions for the past 30 years. Our main findings were: (i) In the face of
agroecological transitions, farmers differ in their management strategies, practices and principles;
(ii) farmers identified as agroecological typically had stronger engagements in a network composed
of farmers’ organisations, universities and NGO’s; (iii) agroecological farms showed great potential
to provide a wide range of ecosystem services as they featured a higher crop diversity and
a higher number of crops for self-consumption; (iv) to promote agroecology, it is crucial to
recognise peasant knowledge, to change the dominant discourse on agriculture through social
movement dynamics, and to generate support from public policies and funds; and (v) participatory
and quantitative methodologies can be combined for more precise and relevant assessments of
agroecological transitions.

Keywords: peasant; agroecological practice; participatory farm typology; statistical farm typology;
participatory action research

Sustainability 2018, 10, 4337; doi:10.3390/su10124337 www.mdpi.com/journal/sustainability

Sustainability 2018, 10, 4337 2 of 20

1. Introduction
Agroecology as an approach to foster the transition to food systems that conserve resources
and improve human well-being [1–3], has been increasingly promoted by scientists [4,5],
Non-Governmental Organisation (NGOs) [6], international organisations [7], and peasant
movements [8]. Agroecology advocates for small-scale, autonomous, resilient and efficient farming
systems, that also value human rights (including women, youth and indigenous people), local cultures,
social participation and food traditions [7]. At the farm level, agroecology favours practices based
on multi-functionality and biodiversity to reduce the dependence on external agrochemical inputs
and to enhance ecological processes [9,10]. Agroecological practices are knowledge intensive and
tailored to local ecological conditions and cultural knowledge [11,12]. Agroecological practices
can be technically oriented, such as composting or biocontrol and/or more socially oriented and
promote, for instance, gender equality or local culture. Policies can advance agroecology by supporting
agroecological research and the development of agroecological practices, and by ensuring that farmers
have access to and security over land, and access to markets that valorise agroecological farmers and
their produce [13–15].
The agroecological transition is defined as the gradual change that farmers undergo to adapt and
move from more conventional towards agroecological farming principles, encompassing technological,
societal, institutional and organisational changes in the food system [16,17]. Agroecological transitions
are often developed within a group of farmers at the community, municipality, regional and/or even
international levels [18], and may influence the ways how farmers organise themselves, value their
traditional culture, and relate to each other and other stakeholders. Although the transition to
agroecology follows general principles, each particular farm has a unique way to adopt and adapt
practices and management strategies. Therefore, the challenges towards agroecological transitions
are not the same for all farmers as farmers differ in objectives and values [19], and are embedded in
different social and ecological contexts [20,21].
To assess the implications of farm diversity for promoting agroecological transitions, two main
challenges need to be addressed. The first challenge refers to a conceptual and empirical understanding
of how to assess the diversity of farmers within transition processes. Earlier studies have sought
to understand the diversity of farmers through the notion of “farming styles” or “farm typologies”,
which distinguish different groups of farmers on the basis of the strategies that they pursue, as well
as farm structural variables [22–25]. Farmers are embedded in different social networks that expose
them to particular discourses, but that also grant farmers access to particular resources, information
on farming practices, and public policies [21,26]. It is not clear how practices, discourses and policies
associated with different farming types favour or hamper agroecological transitions and what types of
interventions could best advance the transition for different farm types. The second challenge refers
to an adequate methodological approach to understanding the relation between farm typologies and
transition processes. Several researchers have argued that classifying and characterizing different types
of farms in a way that is both precise and relevant requires the use of both quantitative and qualitative
indicators and methods [27,28]. Others added that for a precise and relevant characterisation,
researchers must engage with farmers’ own knowledge, values and aspirations [29]. To derive
a meaningful farm typology it is important to use indicators that reflect the farmers’ purposeful
actions, which requires the thorough knowledge of farmers’ intent and perceptions. Participatory
methodologies can be used to convey local knowledge and to stimulate farmers involvement in the
research, increasing the effectiveness and applicability of the results [30,31]. It is not clear, however,
how participatory methodologies, and what combination of quantitative and qualitative methods,
can best be used to understand how farm diversity relates to practices, discourses and policies.
In this study we assess how agroecological practices and principles are associated with different
farm types within a process of agroecological transition. We assess variations between farm types
and describe implications for promoting transitions. We also discuss how and why quantitative
and participatory methodologies can be combined for more precise and relevant assessments to
Sustainability 2018, 10, 4337 3 of 20

understand and promote agroecological transitions. The research questions are: (i) How do farm
types differ in agricultural practices and management? (ii) How conducive are different farm types for
fostering agroecological transitions? (iii) How can different research methodologies contribute to the
understanding of farm diversity to promote agroecological transitions? For this purpose, we developed
and evaluated a farm typology combining participatory and quantitative methodologies, analysing
data collected in the Zona da Mata, Minas Gerais, Brazil. The Zona da Mata is a suitable area to
conduct this research because although different actors were engaged in agroecological transitions for
the past 30 years [12], the region still harbours a wide diversity of farming systems. The agroecological
transition is driven by farmers and their organisations in cooperation and partnership with the
University of Viçosa (UFV), the socio-environmental NGO Centro de Tecnologias Alternativas da Zona
da Mata (CTA-ZM), and other organisations and social movements present in Zona da Mata [12].

2. Methods

2.1. Study Region

The Zona da Mata region is located in the south-eastern part of the state of Minas Gerais, Brazil.
It is part of the Atlantic rainforest biome, considered the 5th biodiversity hotspot in the world, due to
its unique and threatened fauna and flora [32]. The area is mountainous, leading to a diverse set
of bio-physical conditions, limiting mechanisation and favouring small scale farming. The average
temperature in the region is 19 ◦ C, average precipitation is 1300 mm, with 3–4 dry months per year.
Altitude ranges from 200 to 1800 m [33]. The main soil type is classified as Oxisol, which is highly
weathered, deep, well-drained and acidic.
The region was populated by native indigenous communities, who were colonised and nearly
extinguished with the arrival of the Portuguese crown in the eighteenth century. After the colonisation,
at the beginning of the nineteenth century, the land was divided among aristocratic landlords from
other parts of the country. The farms were stablished based on the slave labour of black Africans, and
coffee cultivation became the main cash crop in the region. When slavery was abolished in 1888, lack of
labour became a major problem on the farms [34]. Another problem was the decline of soil fertility,
which resulted from many years of coffee production under poor management conditions with low
adoption of conservationist practices and low nutrient input. At the beginning of the twentieth century,
coffee production started to decline. Degraded coffee plantations were replaced by pastures, leading to
great areas of pastures dominated by grasses, such as Brachiaria spp. and Melinis minutiflora. Since there
were no more slaves, landlords were forced to contract labour. This was done mainly in two ways:
Hiring rural workers per day of work or signing a contract with sharecroppers. Sharecroppers were
allowed to live and work on a portion of land, and in compensation the agricultural production had to
be divided with the landlord. The workers used to live in poor conditions, were subjected to the orders
of the landlords and had no autonomy. Within decades, diverse social dynamics in the region led to
the division of land, mainly through heritage and land acquisition by rural workers [35]. Nowadays,
family farms represent around 82% of the total number of farms in the region while more than a half of
the land is owned by medium and large-scale farmers [36], and sharecropping is still common. To be
considered a family farm according to the Brazilian law, the following criteria must be met: (i) Farm
size is less than 96 ha of land; (ii) farm labour and management is predominantly provided by family
members; and (iii) the main family income comes from rural activities [37]. The mountainous relief
and current predominance of family farmers led to a diverse landscape in Zona da Mata, which can
be understood as a dynamic mosaic of land uses, dominated by pastures, coffee fields and secondary
forest patches.
The study is conducted in three municipalities of Zona da Mata: Divino, Araponga and Espera
Feliz. These municipalities have a strong tradition with coffee production and are part of a long-term
participatory process of agroecological transition, which started in the 1970s [12]. The municipalities
Sustainability
Sustainability2018,
2018,10,
10,x 4337
FOR PEER REVIEW 4 of 2020
4 of

alsoconnect
also connecttwo
twoimportant
importantnature
nature conservation
conservation areas: The national
national park
park “Caparaó”
“Caparaó”and
andthe
thestate
statepark
park
“Serra do Brigadeiro” (Figure 1).
“Serra do Brigadeiro” (Figure 1).

Map
Figure1.1.Map
Figure ofof studied
studied municipalities
municipalities in Zona
in Zona da da Mata,
Mata, Minas
Minas Gerais,
Gerais, Brazil.
Brazil. The The
two two protected
protected areas
ofareas of the are
the region region
alsoare also shown.
shown.

2.2. Farm Typology Construction

2.2. Farm Typology Construction
A farm typology was constructed using two complementary methods: A participatory typology
A farm typology was constructed using two complementary methods: A participatory typology
through workshops with farmers in the three municipalities, and a quantitative analysis based on
through workshops with farmers in the three municipalities, and a quantitative analysis based on the
the characterisation of farming systems. The participatory typology was used to gain insights on
characterisation of farming systems. The participatory typology was used to gain insights on farmers’
farmers’ perceptions and knowledge, and to generate hypotheses on family farm diversity in Zona
perceptions and knowledge, and to generate hypotheses on family farm diversity in Zona da Mata
da Mata region. The participatory-based hypotheses were subsequently tested in the quantitative
region. The participatory-based hypotheses were subsequently tested in the quantitative farm
farm characterisation, which provided information on household characteristics, production strategies,
characterisation, which provided information on household characteristics, production strategies, land
land use, participation in public policies and extension services.
use, participation in public policies and extension services.
2.2.1. Participatory Typology
2.2.1. Participatory Typology
The participatory typology was informed by three workshops, one in each municipality (Figure 2).
TheseTheworkshops
participatory typology
were embeddedwas informed by three
in a long-term workshops,
action researchone in each
process in municipality
the Zona da (Figure
Mata
2).region
These[38].
workshops were embedded in a long-term action research process in the
The workshops involved young to old local farmers and representatives of local Zona da Mata
region [38]. The
organisations. Theworkshops involvedparticipating
number of farmers young to old local
in each farmers
of the and representatives
three workshops of local
was 18, 20 and 34,
organisations. The number of farmers participating in each of the three workshops was 18, 20 and 34,
respectively, for a total of 72 participants. During the workshops, farmers were divided into two or three
respectively, for a total of 72 participants. During the workshops, farmers were divided into two or
groups of 6–15 persons. The process in each group was facilitated by one or two researchers. During
three groups of 6–15 persons. The process in each group was facilitated by one or two researchers.
the workshops, farmers and scientists actively participated in a process of co-creating knowledge,
During the workshops, farmers and scientists actively participated in a process of co-creating
whereby scientists withheld from providing their opinion, but facilitated and systematised the process
knowledge, whereby scientists withheld from providing their opinion, but facilitated and
based on farmers’ perceptions and knowledge.
systematised the process based on farmers’ perceptions and knowledge.
Sustainability
Sustainability2018,
2018,10,
10,x 4337
FOR PEER REVIEW 5 5ofof2020

Figure
Figure2. Stepsofofthethe
2. Steps participatory
participatory methodology
methodology developed
developed in ourwhich
in our study, study,
waswhich
used towas used to
understand
understand farmers’ perspectives
farmers’ perspectives on farmindiversity
on farm diversity the Zonaindathe Zona da Mata.
Mata.

Theindicators
The indicatorsand and criteria
criteria for
for characterising farm diversity
diversity defined
defined during
duringthe theparticipatory
participatory
workshopsinineach
workshops eachmunicipality
municipality were
were recorded, compiled
compiled and
and systematised
systematisedin inaamatrix
matrixtotocharacterise
characterise
a aregional
regionaltypology.
typology.The Theimportance
importance ofof each
each indicator
indicator waswasnotnot weighed,
weighed, although
although some
some indicators
indicators can
can play
play a morea more important
important rolerole than
than otherstotocharacterise
others characterisethethe different
different farmer
farmer types.
types. During
Duringthe the
systematisation,we
systematisation, weonly
onlyselected
selected information
information that was commoncommon to to the
thethree
threemunicipalities.
municipalities.After
Afterthe
the
systematisation,the
systematisation, theresults
resultswere
were presented
presented to to farmers in eacheach municipality
municipality to tovalidate
validatethetheresults
results
obtainedfor
obtained forthe
the region.
region. Finally,
Finally, based
based on on the results
results of
of the
theparticipatory
participatoryworkshops,
workshops,we wecreated
created
hypotheseson
hypotheses onfamily
familyfarm
farmdiversity.
diversity.
Theworkshops
The workshops for fordeveloping
developingthethe farm typology
farm werewere
typology in embedded in an action
in embedded in anresearch
action process,
research
in whichinresearchers
process, participated
which researchers in activities
participated together with
in activities farmers,
together with such as visiting
farmers, such asand working
visiting and
in their fields,
working in theirparticipating in meetings,
fields, participating in workshops and social events.
meetings, workshops Furthermore,
and social we shared the
events. Furthermore, we
knowledge
shared generated by
the knowledge this study
generated by in formal
this studyandin informal
formal and ways, fostering
informal reflections
ways, fostering and discussions
reflections and
and promoting
discussions the exchange
and promoting theofexchange
knowledge. of knowledge.

2.2.2.Farm
2.2.2. FarmCharacterisation
Characterisationand
andQuantitative
QuantitativeTypology
Typology
Thequantitative
The quantitativetypology
typologywas wasbased
based onon aa data
data set
set of
of 115
115 family
familyfarming
farmingsystems
systemsselected
selectedbybythe
the
familyfarmers’
family farmers’unions,
unions,which
whichare areactive
activeinineach
eachofofthe
thethree
threemunicipalities,
municipalities,totocapture
capturethethediversity
diversityof
of family
family farming
farming systems
systems in in
thethe region(Figure
region (Figure1).1).The
The unions
unions approached
approached randomrandom families
familiesofoftheir
their
network, indiscriminately of farmers did or did not participate in the
network, indiscriminately of farmers did or did not participate in the participatory participatory typology workshops.
typology
As the unions
workshops. As are
the responsible
unions are for public services,
responsible for publicsuchservices,
as rural such
retirement,
as rural they have contact
retirement, theywith
have
a great proportion of family farmers in the municipalities. Only family farmers
contact with a great proportion of family farmers in the municipalities. Only family farmers were were selected for
the quantitative
selected analysis, because
for the quantitative analysis,they represent
because they the largestthe
represent proportion of farmersof
largest proportion (82%) in Zona
farmers (82%)dain
Mata [36] and are considered prime targets of agroecological transitions, since they
Zona da Mata [36] and are considered prime targets of agroecological transitions, since they own land, own land, and,
therefore,
and, have have
therefore, more more
autonomy to manage
autonomy the landthe
to manage andland
are more interested
and are to conservetothe
more interested ecological
conserve the
capital based
ecological on local
capital basedandontraditional
local and knowledge [39,40]. Data[39,40].
traditional knowledge collection
Datawas conducted
collection was inconducted
partnershipin
with the local
partnership NGO
with theCTA-ZM
local NGO and researchers
CTA-ZM and in 2015 and 2016in
researchers as 2015
part of a publicly
and 2016 asfunded
part ofassessment
a publicly
to characterise agroecosystems based on functional and structural farm
funded assessment to characterise agroecosystems based on functional and structural variables. The information was
farm variables.
The information was obtained by Participatory Rural Appraisal (PRA) techniques, including semi-
Sustainability 2018, 10, 4337 6 of 20

obtained by Participatory Rural Appraisal (PRA) techniques, including semi-structured interviews,

transect walks on the farm, participatory mapping of the farm and mapping of the gender division of
labour [41,42].
Based on the outcomes of the participatory workshops, data availability and consistency,
we selected seven meaningful quantitative variables to categorise farm households in terms of
farm structural, management and social aspects (Supplementary Materials Table S1). The variables
were: (i) n.policies—number of public policies in which households participated; (ii) n.CTA—number
of extension projects from the local NGO (CTA-ZM) in which households were engaged;
(iii) n.prod—number of crop products produced on the farm; (iv) n.pract—number of agroecological
practices and principles adopted on the farm; (v) age—age of the household head (in years);
(vi) farm.area—farm area (in hectares); and (vii) n.consumed—number of food items produced on
the farm consumed by the family. The variables n.CTA and n.policies are used, respectively, as proxies
for the extend of involvement in social networks and participation in public policies. To assess
the number of adopted agroecological practices (n.pract), we used a checklist of practices that were
previously defined by technicians of the CTA-ZM in collaboration with family farmers of the region
(Supplementary Materials, Table S2).

Multivariate Analysis and Farm Classification

Principal component analysis (PCA) was used to generate non-correlated principal components
(PC’s), and cluster analysis for grouping farms into clusters based on their position on the PCA
axes [22,28,43]. We did not include the variable “number of food items produced on the farm consumed
by the family” in the multivariate analysis, because it was strongly correlated with the variable “number
of crop products” obtained on the farm (correlation coefficient >0.9; Supplementary Materials Figure
S1). The first two axes of the principal component were considered for the subsequent cluster analysis.
These two axes were selected according to the Kaiser criterion, which indicates the selection of any
axis with eigenvalues greater than 1. For the cluster analysis, we used a hierarchical, agglomerative
clustering algorithm using Ward’s method to define the number of groups (k). The final set of
clusters was based on the inherent structure (i.e., the mean value of each variable for each cluster)
aiming at intra-cluster homogeneity and inter-cluster heterogeneity (Supplementary Materials Figure
S2). In addition, after classifying the farms according to the cluster analysis, we developed a binary
recursive partitioning tree classification model to visualise decision rules for predicting the classification
of farms on the different farm types according to the six selected variables.

Characterisation of Different Farm Types

We tested whether farm types defined in the cluster analysis explained variation in the seven
quantitative variables listed earlier. We used one-way ANOVA and Tukey post-doc tests for variables
with normal distributions and homogeneous variances, and the non-parametric Kruskal-Wallis test and
post-doc Dunn test when this was not the case. For proportional data we used Pearson’s Chi-squared
and Fischer tests. p-values are mentioned throughout the manuscript.

Agroecological Practices and Principles

We used neural network analysis to assess the agroecological practices and/or principles
associated with each farm type [43] and that can affect the provision of ecosystem services [44].
The neural network was constructed based on a presence/absence table representing the adoption
(or not) of 25 types of agroecological practices and principles on each farm (Supplementary Materials
Table S2). These 25 practices were taken from the above mentioned checklist. The association of each
of the 25 practices with the farm types were visualised using the software Gephi 0.9.2 [45]. The closer
the distance between each practice to the specific farm type in the neural network, the more affinity
the farm type has to such practice. All the statistical analyses were conducted in R (version 3.3.3).
Sustainability 2018, 10, 4337 7 of 20

3. Results

3.1. Participatory Typology

Six main farmer types were identified during the participatory workshops: Agroecological family
farmers, traditional family farmers, conventional family farmers, part-time farmers, large scale farmers
and landless farmers (Table 1 and Supplementary Materials Table S3). Based on the participatory
scoring exercise, 54.1% of the farmers are considered family farmers that own land, 22.6% landless
farmers, 13.3% large-scale farmers and 10% part-time farmers. Among the family farmers, 19.6%
of the farmers are conventional family farmers, 18.3% are traditional family farmers and 16.2% are
agroecological family farmers.
Agroecological family farmers have a high level of autonomy and a strong connection with nature.
They have diversified production systems, make little use of external inputs, do not use pesticides, use
landrace seeds and adopt innovative farming practices. Agroecological farmers are part of a social
network that includes farmers’ organisations, religion-based organisations, universities and NGO’s.
Traditional family farmers still hold peasant characteristics, such as the use of landrace seeds, cultivation
of diversified home gardens for self-consumption and rely on family labour, which contributes to their
autonomy and food sovereignty. They often make use of traditional practices considered, nowadays,
harmful to the environment, such as fire and intensive weeding. Traditional farmers make little
or moderate use of pesticides and chemical fertilisers, especially in coffee systems. Conventional
family farmers have a strong focus on coffee production. Conventional farms are less diversified than
agroecological or traditional family farms, and strongly rely on pesticides and chemical fertilisers.
A large proportion of food for household consumption is often purchased. Conventional farmers
were early adopters of ’Green Revolution’ technologies and are often part of a network consisting
of suppliers of chemical inputs and government extension services. Part-time farmers have off-farm
jobs besides their farming activities. Their main farming activity is coffee production for the market,
which may be combined with the production of other crops. Their parallel jobs are often related to
farming activities, such as commercialisation of agricultural inputs and cash crops. Part-time farmers
can have a strong opinion on the benefits of agrochemical inputs and were found to promote the Green
Revolution discourse amongst family farmers. Large-scale farmers have large farm areas, high coffee
productivity with the high use of external inputs and hired labour. Both large-scale and part-time
farmers adopt practices that can be considered conventional, such as intensive use of pesticides,
fertilisers and monocultures. However, they may adopt some agroecological practices, as long as they
are economically feasible and profitable, such as the use of manure, mowing instead of intensive uproot
weeding and introduction of one or two tree species in the coffee fields and pastures. Large-scale and
part-time farmers often set themselves as examples of the wealth that can be attained when following
the principles of intensive agriculture. Together, they can also control parts of the coffee supply chain
and hold high positions in the public administration of the municipality. Landless farmers do not own
land and work as land labourers on other farms. They are hired by landlords and therefore do not
have any autonomy with regards to land management decisions.
The participatory typology and associated discussion yielded four hypotheses on family farm
diversity, which were further explored in the quantitative analysis: (i) Agroecological family farms
have higher crop diversification, higher food-security, higher adoption of organic farming practices
and higher social participation than conventional and traditional family farms; (ii) The household
head of traditional family farms is older than in conventional and agroecological family farms; (iii)
Agroecological, traditional and conventional family farms do not differ in terms of farm size; and (iv)
Traditional family farms have higher crop diversification and higher food-security than conventional
family farms, but lower than agroecological family farms. These hypotheses were developed to be
further tested and explored in the quantitative typology.
Sustainability 2018, 10, 4337 8 of 20

Table 1. Overview of the consolidated criteria and indicators for characterizing the six different farm
types based on the information obtained in the three municipalities during the participatory workshops
for constructing the typology. Relative strength for each criterion, according to the farm type: + (small);
+ + (medium); + + + (large); NR (not relevant for the specific farm type).

Agroecological Traditional Conventional

Part-Time Large-Scale Landless
Family Family Family
Farmers Farmers Farmers
Farmers Farmers Farmers
Size of the farm + + + + +++ NR
External inputs + +/++ ++/+++ ++/+++ +++ +++ 1
Organic inputs +++ + + + + +1
Equipment and
+ + +/++ +/++ +++ +++ 1
machinery
Market-oriented + + ++ +++ +++ +++ 1
Distance to town NR NR NR + NR NR
Age of the main
NR +++ NR NR NR NR
householder
off-farm income + + + +++ + +
Resource
+/++ +/++ +/++ +/+++ +++ +
endowment
Land ownership + + + + + NR
Hired labour + + +/++ ++ +++ NR
Family labour +++ +++ +++ + + +++
Participation in
+++ + + + + +
social organisations
Forest conservation ++ + ++ + +/++ +/++ 1
Pastures with trees ++ + + + + +1
Grazing intensity + + +/++ +++ +++ +++ 1
Intensive coffee
+ ++ +/++ ++ ++ ++ 1
weeding
Coffee with trees +++ + + + + +1
Presence of
+++ +++ ++ + + +/++
homegarden
Crop diversification +++ ++ + + + +/++
Indigenous local
+++ +++ ++ + + +/++/+++
knowledge
1 In most cases landless farmers are subjected to the orders of the land owners.

3.2. Quantitative Typology

3.2.1. Multivariate Analysis and Farm Type Classification

The quantitative typology was based on a dataset comprising 115 full time family farmers owning
land. Farmers were selected by the farmers’ unions present in each municipality. Therefore, the other
farmer types identified in the participatory exercise (i.e., large-scale, part-time and landless farmers)
were not included. The first two axes of the PCA combined 48.1% of the variance in the dataset
(Figure 3 and Supplementary Materials Figure S3), reflecting the inherent variability of farming
systems. The variables number of public policies, number of CTA projects, number of crop products
and number of agroecological practices were strongly negatively correlated (R > 0.5) with the first
principal component (PC 1), while age of household head and farm area were strongly positively
correlated (R > 0.5) with the second component (PC 2) (Supplementary Materials Table S4). The cluster
analysis based on the first two PC axes grouped family farms in three clusters (Figure 3). The first
cluster match agroecological farms (n = 35; 30.4%), with a positive correlation with number of crop
products, adoption of agroecological practices and participation in extension projects and public
policies (hypothesis i). The second cluster can be interpreted as traditional family farms (n = 17; 14.8%),
 Sustainability 2018, 10, 4337 9 of 20

with a relatively
Sustainability high
2018, 10, x FOR ageREVIEW
PEER of the
household head and positioned in between the other clusters 9 of 20 in
the horizontal axis (hypotheses ii and iv). The third cluster match with conventional family farms
(n = 63; 54.8%; Figure 3). The participatory-based hypotheses (i), (ii) and (iv) were confirmed by the
multivariate
multivariateanalysis. However,
analysis. However, thethe
hypothesis
hypothesisthat allall
that family
familyfarm
farmtypes
typeshad
hadsimilar
similarfarm
farmsize
size (iii)
(iii)
was rejected,
was since
rejected, large
since farm
large size
farm was
size strongly
was correlated
strongly correlated with traditional
with farms.
traditional farms.

Figure 3. PCA output based on a dataset of 115 family farms. Colours indicate different farm
Figure 3. PCA output based on a dataset of 115 family farms. Colours indicate different farm types
types according to the cluster analysis classification (see text and Figure 4 for more explanation).
according to the cluster analysis classification (see text and Figure 4 for more explanation). n.policies—
n.policies—number of public policies in which households participated; n.CTA—number of extension
number of public policies in which households participated; n.CTA—number of extension projects
projects from the local NGO (CTA) in which households were engaged; n.prod—number of crop
from the local
products NGO on
produced (CTA) in which
the farm; households ofwere
n.pract—number engaged; n.prod—number
agroecological of cropadopted
practices and principles products
on
produced on the farm; n.pract—number of agroecological practices and principles adopted
the farm. Age—age of the household head, in years; farm.area—farm area, in hectares. on the farm.
Age—age of the household head, in years; farm.area—farm area, in hectares.
The agglomerative nesting process used to cluster the 115 farms (Figure 3) shows that, although
theThe agglomerative
level of dissimilarity nesting processbetween
(or distance) used to thecluster the was
clusters 115 farms (Figure
quite high, 3) main
three showsgroups
that, although
of family
thefarmers
level ofcould
dissimilarity (or distance) between the clusters was quite high, three
be distinguished, in agreement with those identified through the participatory typology main groups of
family farmers
(Table 1). could be distinguished, in agreement with those identified through the participatory
typology (Table
The 1).
classification tree model used four of the six variables included in the analysis to separate the
The classification
three groups of family tree model(Figure
farmers used four of the sixofvariables
4). Number included
crop products in crop
(>23.5 the analysis
productstoperseparate
farm) andthe
three groups ofin
participation family
at leastfarmers
one CTA (Figure
project4).were
Number of crop indicators
the emerging products (>23.5 cropagroecological
to identify products per family
farm)
andfarms,
participation in at least one CTA project were the emerging indicators to
while larger farm area (either >7.75 or >16 hectares) and age of the household head (>38) wereidentify agroecological
family farms, while
the indicators largertofarm
selected area (either
distinguish >7.75 orfamily
traditional >16 hectares)
farms fromandtheageother
of thetwohousehold head (>38)
types. Participation
were the indicators
in public policies andselected
number to ofdistinguish
agroecological traditional
practices family
adoptedfarms
did from
not play thea other
role attwo types.
classifying
Participation
farmers according to the binary tree model; yet these indicators were closely correlated with number at
in public policies and number of agroecological practices adopted did not play a role of
classifying farmers
crop products andaccording to the
participation binary
in CTA tree model;
projects. In some yet thesecharacteristics
cases, indicators were closely correlated
of conventional family
with number
farms and theof two
cropother
products
farm andtypesparticipation
seemed to beinrelatively
CTA projects.
subtle. In
Forsome cases,seven
instance, characteristics
conventional of
conventional family farms and the two other farm types seemed to be relatively
farms had more than 23.5 crop products, but as they did not participate in any CTA project, they were subtle. For instance,
seven conventional
classified farms had
as conventional more
farms. than 23.5five
Moreover, crop products, but
conventional farmsashad
theyareas
did between
not participate in any
the thresholds
CTA project, they were classified as conventional farms. Moreover, five conventional
of 7.75 and 16 hectares, but were not classified as traditional farms because the age of the household farms had areas
between the thresholds
head was less than 38 years.of 7.75 and 16 hectares, but were not classified as traditional farms because the
age of the household head was less than 38 years.
Sustainability
Sustainability2018,
2018,10,
10,x 4337
FOR PEER REVIEW 1010ofof2020

Treemodel
Figure4.4.Tree
Figure model displaying
displaying threshold
threshold values
values for
for the
the classification
classification of
of farm
farmtypes
typesbased
basedononthethe
variables used in the multivariate analysis. The boxes represent the number of farms following (ornot)
variables used in the multivariate analysis. The boxes represent the number of farms following (or not)
theestablished
the establishedcriteria.
criteria.The
Theterminal
terminal nodes
nodes represent
represent the
the final
final farm
farm type
typeclassification
classificationaccording
accordingtotothe
the
tree model. The misclassification error rate of the model is 0.06097 (7/115). CTA stands for
tree model. The misclassification error rate of the model is 0.06097 (7/115). CTA stands for Centre of Centre of
Alternative Technologies of Zona da Mata, a local NGO working with family farmers.
Alternative Technologies of Zona da Mata, a local NGO working with family farmers.
3.2.2. Farm Types Characterisation
3.2.2. Farm Types Characterisation
The proportions of farms that are beneficiaries of public policies (Pearson’s Chi-squared test,
p < The
0.01)proportions of farms
and CTA projects that aretest,
(Fisher’s beneficiaries
p < 0.001)ofwerepublic policies (Pearson’s
significantly different Chi-squared
among farm test, p<
types.
0.01) and CTA projects (Fisher’s test, p < 0.001) were significantly different among
All agroecological family farmers have participated in at least one project from CTA and almost 70% of farm types. All
agroecological
agroecologicalfamily
familyfarmers
farmershave participated
benefitted from atinleast
at least
threeone project
to four from
public CTA and
policies. Onalmost 70% of
the contrary,
agroecological family farmers benefitted from at least three to four public policies.
the majority of conventional and traditional family farmers did not participate in CTA projects and On the contrary,
the
hadmajority of conventional
less access and
to the benefits traditional
of public family
policies. farmers
Almost onedid notofparticipate
third traditional in CTAfarmers
family projectsdid
and
had
notless access tointhe
participate benefits
public of public
policies policies.
and about 60%Almost one third of traditional
of the conventional family farmersfamilyhadfarmers
access did not
to the
participate in public policies and about 60% of the conventional family
benefits from one or two public policies (Supplementary Materials Tables S5 and S6). farmers had access to the
benefits from one or two
Agroecological public
farms grew policies
a larger(Supplementary
number of crops Materials
(ANOVA, Tables S5 and S6).
p < 0.0001), implemented more
Agroecological farms grew a larger number of crops (ANOVA,
agroecological practices and principles (Kruskal-Wallis, p < 0.0001), and consumed p < 0.0001),a greater
implemented more
diversity of
agroecological practices and principles (Kruskal-Wallis, p < 0.0001), and consumed
crops produced on farm (ANOVA, p < 0.0001) than conventional and traditional family farms (Figure 5). a greater diversity
ofTraditional
crops produced on farm (ANOVA,
and conventional family farmsp < 0.0001) than conventional
have, however, adopted some and agroecological
traditional family farms
practices
(Figure 5). Traditional and conventional family farms have, however, adopted some agroecological
and grow a lower, but still considerable, diversity of crops and of products consumed (Figure 5).
practices and grow a lower, but still considerable, diversity of crops and of products consumed
Besides, the household head of agroecological family farms are significantly older (Kruskal-Wallis,
(Figure 5). Besides, the household head of agroecological family farms are significantly older (Kruskal-
p < 0.0001) than conventional family farmers (Figure 5), which suggests that younger farmers may
Wallis, p < 0.0001) than conventional family farmers (Figure 5), which suggests that younger farmers
be more inclined to follow a market-oriented production orientation. Finally, the differences among
may be more inclined to follow a market-oriented production orientation. Finally, the differences
traditional and the other two family farm types indicated by the participatory typology were confirmed
among traditional and the other two family farm types indicated by the participatory typology were
by significantly older household heads (Kruskal-Wallis, p < 0.0001) and larger farm sizes (ANOVA,
confirmed by significantly older household heads (Kruskal-Wallis, p < 0.0001) and larger farm sizes
p < 0.0001) associated with traditional family farmers (Supplementary Materials Tables S7 and S8).
(ANOVA, p < 0.0001) associated with traditional family farmers (Supplementary Materials Tables S7
and S8).
Sustainability
Sustainability2018,
2018,10,
10,x 4337
FOR PEER REVIEW 1111
ofof
2020

Boxplotsofof agroecological,
Figure5.5.Boxplots
Figure agroecological, traditional
traditional and
and conventional
conventional farmers
farmers for
for five
fivefarming
farmingsystem
system
social characteristics. The black line that divides the box represents the median. The end of theboxes
social characteristics. The black line that divides the box represents the median. The end of the boxes
representsfirst
represents firstand
andthird
thirdquartiles,
quartiles, whiskers
whiskers indicate
indicate the
the minimum
minimum and andmaximum
maximumvaluesvalues(excluding
(excluding
outliers)and
outliers) andoutlier
outliervalues
valuesare
are represented
represented separately
separately byby the
the small
small circles.
circles. Mean
Meanvalues
valuesand
andletters
letters
indicating statistically significant differences are indicated above the boxplots. Age the
indicating statistically significant differences are indicated above the boxplots. Age the household headhousehold
head
and and number
number of agroecological
of agroecological practices
practices andand principles
principles wereanalysed
were analysed using
using Kruskal-Wallis,
Kruskal-Wallis,and and
number of crop products consumed, number of crop products and farm
number of crop products consumed, number of crop products and farm area using ANOVA. area using ANOVA.

3.2.3. Association between Farm Diversity and Agroecological Practices and Principles
3.2.3. Association between Farm Diversity and Agroecological Practices and Principles
The neural network analysis showed variable degrees of association between agroecological
The neural network analysis showed variable degrees of association between agroecological
practices and principles and the three types of family farms (Figure 6 and Supplementary Materials
practices and principles and the three types of family farms (Figure 6 and Supplementary Materials
Table S9). Agroecological family farms appeared at the centre of the network, while conventional
Table S9). Agroecological family farms appeared at the centre of the network, while conventional and
and traditional farms had a peripheral position, reflecting the expected general higher association of
traditional farms had a peripheral position, reflecting the expected general higher association of
agroecological farms with all the practices and principles included in the analysis. Practices, such as
agroecological farms with all the practices and principles included in the analysis. Practices, such as
composting, use of bio-fertilisers and alternative pesticides, cultivation of repellent plants, the presence
composting, use of bio-fertilisers and alternative pesticides, cultivation of repellent plants, the
of agroforestry systems, and local culture valorisation appeared on the upper left hand-side of the
presence of agroforestry systems, and local culture valorisation appeared on the upper left hand-side
network. These socio-ecological indicators were more distant from traditional and conventional family
of the network. These socio-ecological indicators were more distant from traditional and conventional
farms and more closely associated with agroecological farms, indicating a more advanced stage of
family farms and more closely associated with agroecological farms, indicating a more advanced stage
agroecological transition. Gender equity and group participation were also more closely associated
of agroecological transition. Gender equity and group participation were also more closely associated
with agroecological farms. There was a set of practices, such as the use of manure, exchange of
with agroecological farms. There was a set of practices, such as the use of manure, exchange of labour
labour days, family collaboration, use of landrace seeds, use and cultivation of medicinal plants and
days, family collaboration, use of landrace seeds, use and cultivation of medicinal plants and
intercropping that were shared across the three family farm types. There were also practices (e.g.,
intercropping that were shared across the three family farm types. There were also practices (e.g.,
mechanical weed mowing or forest conservation) more commonly associated with conventional and
mechanical weed mowing or forest conservation) more commonly associated with conventional and
agroecological family farmers than with traditional farmers. Moreover, alternative animal feeding to
agroecological family farmers than with traditional farmers. Moreover, alternative animal feeding to
complement grazing and presence of on-farm scattered trees were more common among agroecological
complement grazing and presence of on-farm scattered trees were more common among
and traditional family farms than for conventional farms.
agroecological and traditional family farms than for conventional farms.
Sustainability
Sustainability2018,
2018,10,
10,x 4337
FOR PEER REVIEW 1212ofof2020

Neuralnetwork
Figure6.6.Neural
Figure network representing
representing thethe association
association between
between thethe different
different farm
farmtypes
typeswith
witheach
each
agroecological practice and principle. The closer the practice/principle to the specific
agroecological practice and principle. The closer the practice/principle to the specific farm type, farm type,
the
the more
more affinity
affinity the type
the farm farm has
typetohas to practice/principle.
such such practice/principle.
Compost Compost = Composting
= Composting of organic
of organic residues;
residues; repelent_plant
repelent_plant = Presence= of
Presence
plants ofused
plants
to used
repeltopests;
repel onfarm_trees
pests; onfarm_trees = Presence
= Presence of on-farm
of on-farm trees
trees scattered around the house, around or within the fields; biofert = Use of biofertilisers originated
scattered around the house, around or within the fields; biofert = Use of biofertilisers originated from
from organic material; crop_rotation = Presence of crop rotation systems; alt_pesticide = Use of
organic material; crop_rotation = Presence of crop rotation systems; alt_pesticide = Use of alternative
alternative pesticides originated from organic material; AFS = Presence of agroforestry systems;
pesticides originated from organic material; AFS = Presence of agroforestry systems; no_transgenics =
no_transgenics = No use of transgenic seeds and crops; alt_feeding = Alternative on-farm feed
No use of transgenic seeds and crops; alt_feeding = Alternative on-farm feed production (e.g., sugar
production (e.g., sugar cane, tree residues, grasses, banana, fruits); manure = Use of animal manure as
cane, tree residues, grasses, banana, fruits); manure = Use of animal manure as plant fertilisers;
plant fertilisers; diversification = Presence of more than ten crops in the farming system; med_plants
diversification = Presence of more than ten crops in the farming system; med_plants = Cultivation and
= Cultivation and use of medicinal plants; intercropping = Presence of intercropping systems (e.g.,
use of medicinal plants; intercropping = Presence of intercropping systems (e.g., maize and beans);
maize and beans); mowing = Predominance of coastal mechanical mowing for weeding instead
mowing = Predominance of coastal mechanical mowing for weeding instead of uprooting weeds;
of uprooting weeds; no_fire = No use of fire to manage the fields; no_chemicalfert = No use of
no_fire = No use of fire to manage the fields; no_chemicalfert = No use of chemical fertilisers;
chemical fertilisers; race_seeds = Use of at least one variety of local seeds varieties; forest_conserv
race_seeds = Use of at least one variety of local seeds varieties; forest_conserv = Presence and
= Presence and conservation of forest patches on the farm; homeop = Use of homeopathy for
conservation of forest patches on the farm; homeop = Use of homeopathy for humans, animals, soil
humans, animals, soil and/or plants; water_conser = Conservation of water springs present in
and/or plants; water_conser = Conservation of water springs present in the farm; family_colaboration
the farm; family_colaboration = Most family members actively contribute to farm management;
= Most family members actively contribute to farm management; gender_equity = Recognition of the
gender_equity = Recognition of the women’s work importance; group_participation = Participation in
women’s work importance; group_participation = Participation in social organisations, such as farmers
social organisations, such as farmers unions, associations and cooperatives; labour_exchange = Labour
unions, associations and cooperatives; labour_exchange = Labour exchange with other farmers in the
exchange with other farmers in the community; popular_culture = Participation and organisation in
community; popular_culture = Participation and organisation in events to celebrate and reinforce the
events to celebrate and reinforce the local culture.
local culture.
4. Discussion
4. Discussion
The participatory and statistical typologies of farm households in the Zona da Mata distinguished
The participatory
three main and statistical
types of family typologies of farm
farms: Conventional, households
traditional in the Zona da ones.
and agroecological Mata distinguished
We showed
three
that main types
certain of family
practices farms: Conventional,
commonly traditional and
labelled as agroecological wereagroecological
common in allones. We farms
family showed that
(e.g.,
certain practicesand
intercropping commonly labelled while
use of manure), as agroecological weresuch
other indicators, common
as the in all family with
engagement farmssocial
(e.g.,
intercropping and use of manure), while other indicators, such as the engagement with social
Sustainability 2018, 10, 4337 13 of 20

organisations, NGO’s, local culture, and public policies were distinctive features of agroecological
farms. This reinforces the idea that agroecology needs to be defined not only by practices, but also by
principles and social relations.

4.1. Agroecological Practices and Principles

Family farmers differed in terms of farm management and adoption of agroecological practices
and principles, and the association between these and the farm typology was revealed through the
neural network analysis (Figure 6). The use of chemical fertilisers was common in all family farm
types. Agroecological farms also adopted other practices to maintain soil fertility, including the
use of compost, biofertilisers, and agroforestry. These practices reduce the need to apply chemical
fertilisers and generate other benefits for the soil, such as increased organic matter content, reduced
soil erosion and improvement of soil biological and physical quality [46,47]. Alternative practices for
biological control were also more common in agroecological farms, such as the use of plant extracts
to manage pests and diseases, and the cultivation of repellent plants [48,49]. Social capital was also
stronger amongst agroecological farms, which is reflected in higher group participation and gender
equity. Indeed, social capital may contribute to farmers’ ability to self-organise, claim rights and access
knowledge and information [50,51].
Some agroecological practices that have their roots in peasant traditions [13] are still present
amongst family farms. They include intercropping, exchange of labour days, the use of landrace seeds
and the use and cultivation of medicinal plants. Building upon these practices can be a good strategy
for policies and extension services to support a process of agroecological transition, because farmers
are already accustomed to, and have knowledge of them [52]. Alternative animal feeding practices
and the presence of on-farm scattered trees are associated with the management of multi-species
home gardens which are often encountered on traditional and agroecological farms. Home gardens
are often managed without external inputs and supported by biodiversity-mediated processes and
resources produced within the farming system, such as manure and plant residues. This makes farmers
more autonomous in terms of producing food, both for self-consumption and for the market [53,54].
In addition, traditional knowledge about plants and animals is preserved and put in practice in home
gardens [55]. Women manage home gardens and therefore play an important role in maintaining
ancestral knowledge related to agroecological practices [56].
According to the participatory typology, the use of fire and intensive weeding were more common
in traditional farms. Before the Green Revolution, these practices were considered beneficial for crop
production, and fire was applied to reduce labour requirements [57]. Nowadays, some agroecological
practices are also broadly adopted in conventional farms (e.g., mechanical mowing instead of uproot
weeding), which can generate the same benefits as fire and intensive weeding, while reducing
soil erosion and improving nutrient cycling and water conservation. Furthermore, conventional
family farmers were more likely to conserve forest patches on their farms than agroecological and
traditional farmers, probably because they acquired land that already contained forest. While currently
there are strict laws that can penalise farmers for deforestation, family farmers are typically not
motivated to restore new forest areas. Reforestation is a challenge for family farmers because of
reduced land availability and little knowledge on restoration strategies and their possible benefits.
While environmental laws are needed to maintain the reminiscent forest patches, these currently do
not provide efficient incentives for smallholder farmers to restore land by managing native vegetation.
For instance, farmers are discouraged to stimulate regrowth of native vegetation because they can
be punished in case they need to cut or manage the trees in the future [58]. This indicates a need
for (i) adjusted policies that better incentivise farmers to conserve native vegetation, in combination
with (ii) an engagement process with farmers to understand and strengthen multiple ecosystem
services derived from the forest, and (iii) the development of native vegetation management and
conservation strategies that fit farmer’s realities through extension programs and activities from the
agroecological movement.
Sustainability 2018, 10, 4337 14 of 20

4.2. Farm Diversity and Implications for Agroecological Transitions

Farming systems in the Zona da Mata are inherently complex and diverse (Figures 5 and 6).
Despite the aim of typology approaches to identify discrete groups, in reality farm diversity can best
be understood as a continuum where different farm types can co-evolve, interact and overlap [59].
In fact, agroecological transitions may also be understood as a process in which farmers move along an
infinite continuum, and it is therefore difficult to draw a sharp line that separates agroecological from
non-agroecological farmers, as well as a specific end point of transition. In our study, this is evidenced
by the fuzzy boundaries between conventional, traditional and agroecological farms (Figures 4–6).
Indeed, there was great variation within each farm type, for instance in terms of the wide set of practices
adopted by farmers in the same type. Nevertheless, farm types can be understood as reference points
in the same continuum and our typology was able to make explicit the diversity of farming systems,
and to reveal clear and useful general patterns that provide insights on how to assess and foster
transition processes.
The quantitative typology showed that farms with converging characteristics that reflect high
agrobiodiversity, food security and social engagement were statistically clustered in one type
(Figures 3 and 6). Therefore, we can consider these farms to be more advanced in the agroecological
transition process and to have a presumably high potential for the provision of a wide range of
ecosystem services [60]. By engaging with social organisations, family farmers may be able to share and
access knowledge and participate in interactions with other stakeholders which can foster the transition
towards more agroecological systems [61,62]. In Zona da Mata, farmers categorised as agroecological,
who in the past were either landless, conventional or traditional family farmers, were part of a social
movement and a network that brought together farmers, church-based organisations, local NGOs
(CTA), the Federal University of Viçosa, and other organisations [12] (Table 1 and Figure 3). Through
this movement, they were able to contest the mainstream discourse of the Green Revolution and
construct an alternative agroecological discourse that recognises farmers’ own knowledge and
resources [11]. Through this movement, traditional practices were revalued and new agroecological
practices constructed through on-farm experiments, participatory research and peasant-to-peasant
learning exchanges [12,31].
The movement also allowed farmers to participate and co-construct a variety of public policies,
including the Land Credit Policy, the Food Acquisition Programme and the National School Feeding
Program (Figure 3). These policies enabled farmers to access land and to create local markets that
gave more value to agroecological products [63]. The maintenance and development of adequate
policies and funds to better support more sustainable and equitable food systems could accelerate the
expansion of agroecological networks and activities [64]. For instance, the significant lower age of
the household head of conventional compared to agroecological family farmers (Figure 5) indicates
the need to support young farmers in the process of transitions, providing them access to credit and
ecologically-based extension services.
Amongst family farms, the traditional farms were led by older household heads that tended to
hold larger areas of land (Figure 5), as the heritage process resulting in the division of the land to their
heirs was not yet concluded. In our study the average farm size of family farms was 7.3 ha, while data
from the national census show that the average farm size in the three municipalities is 9.3 ha for family
farmers and 47.10 ha for non-family farmers [36]. Our results suggest that traditional family farmers
could be an important target of agroecological transitions as they have heirs who will take their place
in the farm and still hold many peasant characteristics that are supportive of agroecology, such as
indigenous knowledge, strong community networks, high agrobiodiversity, conservation of landrace
seeds and food self-sufficiency [65] (Table 1, Figures 5 and 6).
Both part-time and large scale farmers appeared as a less suitable target for agroecological
transitions, because of their view on nature as a commodity [19], as well as strong market-orientation
and economical focus. However, large scale and part-time farmers can still be motivated to adopt
agroecological practices that have no negative impact on cash crop production. For instance, reducing
Sustainability 2018, 10, 4337 15 of 20

the use of chemical pesticides, mowing instead of intensive weeding, abandoning the use of fire and
implementing water harvesting structures may be considered interesting by more market-oriented
farmers. However, adoption of these practices will only result in a small transition shift, as it does not
include social, cultural and organisational aspects that are also part of agroecology [66].
Farmers only made significant advances towards agroecology once they had legal land rights [67].
Several agroecological farmers acquired land through the government’s Land Credit Policy [68]
and through locally organised arrangements that purchase land collectively [69]. The importance of
policies and movements that promote land rights, such as the Land Credit Policy, need to be re-affirmed
nationally and internationally, as these policies are being dismantled and movements criminalised
by the current government. The landless movement is also present in the Zona da Mata region and
requires support as a way to promote an agrarian reform. Brazil is one of the countries with the
greatest land concentration in the world which is associated with negative externalities, such as rural
poverty and social exclusion [68]. Critically, access to land should be followed by programmes that
allow farmers to exchange and acquire knowledge of agroecological farming. Otherwise there is a
risk that these farmers will follow the principles set by the dominant discourse on market-oriented
agriculture [70].

4.3. Typology Construction for Agroecological Transitions

To our knowledge, this is the first study that developed farm typologies specifically to understand
and analyse a long-term process of agroecological transition, focusing on changes at farm level.
Following previous research in the region [12,38], we used participatory methods to interest farmers
to participate in a collective process of co-creation of knowledge [31]. This was also important to
generate a collective understanding of agroecology. This collective understanding is relevant to increase
awareness about agroecological ideas, farms and practices, as well as to identify opportunities and
barriers for promoting agroecological transitions [71]. Participating farmers reported that the process
allowed them to better understand and reflect upon their own reality, which may have implications
for their future management choices. The outcomes of the participatory typology did not always
match with empirical statistics. For instance, the proportion of family farmers in Zona da Mata in
our study (54.1%) was lower than reported in the national census (82%). This discrepancy can be
explained by the fact that landless farmers were considered as a separate category in our participatory
workshops and not in the national census. The number of landless farmers in the national census is
also probably underestimated, due to difficulties to interview rural workers who do not own land.
While our participatory approach allowed the collection of detailed context specific data, it was based
on the perceptions of farmers, which does not necessarily accurately represent the reality.
The participatory typology offers a qualitative, broad and holistic view on farm diversity,
typical of indigenous knowledge systems [72], while the statistical typology takes a relatively
reductionist approach based on a few variables. Nevertheless, the quantitative analysis allowed
the further exploration and testing of the general patterns that emerged from the participatory
typology. The outcomes of the quantitative analysis brought insights about the relationships between
farm structural and social variables and the position of farmers on a transition continuum among
conventional, traditional and agroecological systems. Although the tree classification model provided
useful quantitative information on thresholds to be used as reference values for further categorisation
and/or monitoring of agroecological transitions, it should be examined carefully, as farm types featured
great variation and fuzzy boundaries. The only major discrepancy between the categorisation made by
the two approaches was the farm size among family farmers. In the quantitative typology, farm size
was higher in traditional than in the two other family farm types. In contrast, no differences on farm size
among family farmers were shown in the participatory workshops, probably because the differences
between family and large-scale farms were much more prominent, or because farmers do not perceive
this variable as relevant. We also suggest that quantitative variables that reflect participation in social
networks must be considered in future studies on agroecological transitions. In our case, although the
Sustainability 2018, 10, 4337 16 of 20

variable “n.CTA—number of extension projects from CTA in which households were engaged” was
endogenous to our own data set, it represents the engagement of farmers in a network, or a movement.
Although our effort is a first step towards identifying and developing indicators that account for social
and network engagement, we recognise the limitation of the variable “n.CTA” as there are different
ways and degrees for participating in a movement. The level of participation or engagement needs to
be better captured as we have shown that it plays an important role in agroecological transitions.

5. Conclusions
We combined participatory and quantitative methodologies to highlight contrasting characteristics
among farm types, including differences in terms of farmers’ agroecological practices and principles.
Our study revealed how farmers perceived the different types of farms in the region and what are
the implications of farm diversity for the strategies to promote agroecological transitions. Our main
findings are:

(i) In the face of agroecological transitions, farmers differ in their management strategies, practices
and principles;
(ii) Farmers identified as agroecological were strongly engaged in a network composed of farmers’
organisations, universities and NGO’s;
(iii) Agroecological farms showed great potential to provide a wide range of ecosystem services as
they featured higher crop diversity and higher number of crops for self-consumption;
(iv) To promote agroecology, it is crucial to recognise peasant knowledge and change the dominant
discourse on agriculture through social movement dynamics, as well as to generate support from
public policies and funds; and
(v) Participatory and quantitative methodologies can be combined for more precise and relevant
assessments of agroecological transitions

To further assess why and how each agroecological practice is adopted, indicators of the intensity
of adoption need to be developed. Future work should address how agroecological practices
and principles are constructed and disseminated, considering infra and supra-household drivers,
interconnections among social and ecological variables, as well as the influence of organisations,
local culture and knowledge. The promotion of agroecological transitions must place farmers and their
knowledge at the centre. In addition, farmer organisations, scientists, policy-makers and technicians
must get together to facilitate knowledge exchange and dissemination, as well as to incorporate
agroecology in national and international political agendas towards more equitable, sustainable and
autonomous agriculture and food systems.

Supplementary Materials: The following are available online at http://www.mdpi.com/2071-1050/10/12/4337/s1.

Author Contributions: Conceptualisation, H.M.T., I.M.C., A.J.V. and P.T.; Data curation, H.M.T. and A.J.V.; Formal
analysis, H.M.T.; Investigation, H.M.T., L.v.d.B., I.M.C. and A.J.V.; Methodology, H.M.T., L.v.d.B., I.M.C. and A.J.V.;
Project administration, H.M.T. and I.M.C.; Resources, H.M.T. and I.M.C.; Supervision, I.M.C., F.J.J.A.B., M.P.-C.
and P.T.; Writing—original draft, H.M.T.; Writing—review and editing, H.M.T., L.v.d.B., I.M.C., A.J.V., F.J.J.A.B.,
M.P.-C. and P.T.
Funding: This research was funded by the Interdisciplinary Research and Education Fund of Wageningen
University (INREF); Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG): APQ-03348-16;
Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq): Universal 01/2016, and the Ministry of
Agrarian Development of the Brazilian government (MDA): Public Call nº13/2013. The APC was funded by the
Farming Systems Ecology chairgroup of Wageningen University.
Acknowledgments: We would like to thank farmers, farmer organisations’, CTA-ZM and the agroecology
movement in Zona da Mata for making this research possible. This research is part of the FOREFRONT programme,
funded by the Interdisciplinary Research and Education Fund (INREF) of Wageningen University and Research.
We thank and acknowledge the contributions of all members of the FOREFRONT program, forming together a
truly interdisciplinary research team. We thank FAPEMIG, CNPq and the Ministry of Agrarian Development of
the former federal Brazilian government for the funding support.
Conflicts of Interest: The authors declare no conflict of interest.
Sustainability 2018, 10, 4337 17 of 20

References
1. Dupré, M.; Michels, T.; Gal, P. Le Diverse dynamics in agroecological transitions on fruit tree farms.
Eur. J. Agron. 2017, 90, 23–33. [CrossRef]
2. Blesh, J.; Wolf, S.A. Transitions to agroecological farming systems in the Mississippi River Basin: Toward an
integrated socioecological analysis. Agric. Hum. Values 2014, 31, 621–635. [CrossRef]
3. Beudou, J.; Martin, G.; Ryschawy, J. Cultural and territorial vitality services play a key role in livestock
agroecological transition in France. Agron. Sustain. Dev. 2017, 37, 1–11. [CrossRef]
4. Nicholls, C.I.; Altieri, M.A. Pathways for the amplification of agroecology. Agroecol. Sustain. Food Syst. 2018,
1–24. [CrossRef]
5. Silici, L. Agroecology: What It Is and What It Has to Offer; IIED: London, UK, 2014.
6. Oxfam. Building a New Agricultural Future; Oxfam: Cambridge, UK, 2014; pp. 1–18.
7. FAO. FAO’s Work on Agroecology: A Pathway to Achieving the SDG’s; FAO: Rome, Italy, 2018.
8. La Via Campesina. Peasant Agroecology for Food Sovereignty and Mother Earth; La Via Campesina: Mons,
Belgium, 2015.
9. Duru, M.; Therond, O.; Fares, M. Designing agroecological transitions; A review. Agron. Sustain. Dev. 2015,
35, 1237–1257. [CrossRef]
10. Bonaudo, T.; Bendahan, A.B.; Sabatier, R.; Ryschawy, J.; Bellon, S.; Leger, F.; Magda, D.; Tichit, M.
Agroecological principles for the redesign of integrated crop-livestock systems. Eur. J. Agron. 2014, 57, 43–51.
[CrossRef]
11. Van den Berg, L.; Roep, D.; Hebinck, P.; Teixeira, H.M. Reassembling nature and culture: Resourceful farming
in Araponga, Brazil. J. Rural Stud. 2018, 61, 314–322. [CrossRef]
12. Cardoso, I.M.; Guijt, I.; Franco, F.S.; Carvalho, A.F.; Ferreira Neto, P.S. Continual learning for agroforestry
system design: University, NGO and farmer partnership in Minas Gerais, Brazil. Agric. Syst. 2001, 69,
235–257. [CrossRef]
13. Copeland, N. Meeting peasants where they are: Cultivating agroecological alternatives in neoliberal
Guatemala. J. Peasant Stud. 2018, 1–22. [CrossRef]
14. Mccune, N.; Rosset, P.M.; Salazar, T.C.; Moreno, A.S.; Morales, H. Mediated territoriality: Rural workers and
the efforts to scale out agroecology in Nicaragua. J. Peasant Stud. 2017, 44, 354–376. [CrossRef]
15. Kepkiewicz, L.; Dale, B.; Kepkiewicz, L.; Dale, B. Keeping “our” land: Property, agriculture and tensions
between Indigenous and settler visions of food sovereignty in Canada in Canada. J. Peasant Stud. 2018, 1–20.
[CrossRef]
16. Tittonell, P. Ecological intensification of agriculture-sustainable by nature. Environ. Sustain. 2014, 8, 53–61.
[CrossRef]
17. Gliessman, S. Transforming food systems with agroecology. Agroecol. Sustain. Food Syst. 2016, 40, 187–189.
[CrossRef]
18. Guzmán, G.I.; López, D.; Román, L.; Alonso, A.M. Participatory Action research in agroecology: Building
local organic food networks in Spain. Agroecol. Sustain. Food Syst. 2013, 37, 127–146. [CrossRef]
19. Teixeira, H.M.; Vermue, A.J.; Cardoso, I.M.; Peña Claros, M.; Bianchi, F.J.J.A. Farmers show complex and
contrasting perceptions on ecosystem services and their management. Ecosyst. Serv. 2018, 33, 44–58.
[CrossRef]
20. Leeuwis, C.; Pyburn, R. Wheelbarrow Full of Frogs: Social Learning in Rural Resource Management; Van Gorcum
Ltd.: Assen, The Netherlands, 2002; ISBN 9023238508.
21. Weltin, M.; Zasada, I.; Franke, C.; Piorr, A.; Raggi, M.; Viaggi, D. Analysing behavioural differences of
farm households: An example of income diversification strategies based on European farm survey data.
Land Use Policy 2017, 62, 172–184. [CrossRef]
22. Kuivanen, K.S.; Alvarez, S.; Michalscheck, M.; Adjei-Nsiah, S.; Descheemaeker, K. Characterising the
diversity of smallholder farming systems and their constraints and opportunities for innovation: A case
study from the Northern Region, Ghana. NJAS—Wagen. J. Life Sci. 2016, 78, 153–166. [CrossRef]
23. Cortez-Arriola, J.; Rossing, W.A.H.; Massiotti, R.D.A.; Scholberg, J.M.S.; Groot, J.C.J.; Tittonell, P. Leverages
for on-farm innovation from farm typologies? An illustration for family-based dairy farms in north-west.
Agric. Syst. 2015, 135, 66–76. [CrossRef]
Sustainability 2018, 10, 4337 18 of 20

24. Schmitzberger, I.; Wrbka, T.; Steurer, B.; Aschenbrenner, G.; Peterseil, J.; Zechmeister, H.G. How farming
styles influence biodiversity maintenance in Austrian agricultural landscapes. Agric. Ecosyst. Environ. 2005,
108, 274–290. [CrossRef]
25. Kansiime, M.K.; van Asten, P.; Sneyers, K. Farm diversity and resource use efficiency: Targeting agricultural
policy interventions in East Africa farming systems. NJAS—Wagen. J. Life Sci. 2018, 85, 32–41. [CrossRef]
26. Van der Ploeg, J.D. The New Peasantries: Struggles for Autonomy and Sustainability in an Era of Empire and
Globalization, 1st ed.; Earthscan: London, UK, 2008; ISBN 1844078825.
27. Righi, E.; Dogliotti, S.; Stefanini, F.M.; Pacini, G.C. Capturing farm diversity at regional level to up-scale
farm level impact assessment of sustainable development options. Agric. Ecosyst. Environ. 2011, 142, 63–74.
[CrossRef]
28. Modernel, P.; Dogliotti, S.; Alvarez, S.; Corbeels, M.; Picasso, V.; Tittonell, P.; Rossing, W.A.H. Identification
of beef production farms in the Pampas and Campos area that stand out in economic and environmental
performance. Ecol. Indic. 2018, 89, 755–770. [CrossRef]
29. Leeuwis, C.; Van den Ban, A. Communication for Rural Innovation, 3rd ed.; Blackwell Publishing: London, UK,
2004; ISBN 9780470995235.
30. Kuivanen, K.S.; Michalscheck, M.; Descheemaeker, K.; Adjei-Nsiah, S.; Mellon-Bedi, S.; Groot, J.C.J.;
Alvarez, S. A comparison of statistical and participatory clustering of smallholder farming systems—A case
study in Northern Ghana. J. Rural Stud. 2016, 45, 184–198. [CrossRef]
31. Méndez, V.E.; Caswell, M.; Gliessman, S.R.; Cohen, R. Integrating agroecology and Participatory Action
Research (PAR): Lessons from Central America. Sustainability 2017, 9, 705. [CrossRef]
32. Myers, N.; Mittermeier, R.A.; Fonseca, G.A.B.; Fonseca, G.A.B.; Kent, J. Biodiversity hotspots for conservation
priorities. Nature 2000, 403, 853–858. [CrossRef] [PubMed]
33. Golfari, L. Zoneamento Ecológico do Estado de Minas Gerais Para Reflorestamento; Centro de Pesquisa Florestal
da Região do Cerrado: Belo Horizonte, Brazil, 1975.
34. Freire, J. Escravidão e Família Escrava Na Zona da Mata Mineira Oitocentista. Ph.D. Thesis, Universidade
Estadual de Campinas, Campinas, Brazil, 2009.
35. Valverde, O. Estudo regional da Zona da Mata de Minas Gerais. Rev. Bras. Geogr. 1958, 1, 1–131.
36. IBGE Instituto Brasileiro De Geografia E Estatística. Censo Agropecuário 2006; IBGE: Rio de Janeiro, Brazil,
2006.
37. LEI N◦ 11.326. Brazil National Policy on Family Farming and Rural Family Enterprises 2006. Available
online: http://www2.camara.leg.br/legin/fed/lei/2006/lei-11326-24-julho-2006-544830-normaatualizada-
pl.html (accessed on 15 February 2018).
38. Souza, H.N.; Cardoso, I.M.; de Sá Mendonça, E.; Carvalho, A.F.; de Oliveira, G.B.; Gjorup, D.F.; Bonfim, V.R.
Learning by doing: A participatory methodology for systematization of experiments with agroforestry
systems, with an example of its application. Agrofor. Syst. 2012, 85, 247–262. [CrossRef]
39. Bernard, B.; Lux, A. How to feed the world sustainably: An overview of the discourse on agroecology and
sustainable intensification. Reg. Environ. Chang. 2017, 17, 1279–1290. [CrossRef]
40. Van der Ploeg, J.D. Peasant-driven agricultural growth and food sovereignty. J. Peasant Stud. 2014, 41,
999–1030. [CrossRef]
41. Chambers, R. The origins and practice of participatory rural appraisal. World Dev. 1994, 22, 953–969.
[CrossRef]
42. Tittonell, P.; Vanlauwe, B.; Leffelaar, P.A.; Rowe, E.C.; Giller, K.E. Exploring diversity in soil fertility
management of smallholder farms in western Kenya: I. Heterogeneity at region and farm scale.
Agric. Ecosyst. Environ. 2005, 110, 149–165. [CrossRef]
43. Tittonell, P.; Muriuki, A.; Shepherd, K.D.; Mugendi, D.; Kaizzi, K.C.; Okeyo, J.; Verchot, L.; Coe, R.;
Vanlauwe, B. The diversity of rural livelihoods and their influence on soil fertility in agricultural systems of
East Africa–A typology of smallholder farms. Agric. Syst. 2010, 103, 83–97. [CrossRef]
44. Mulder, C.; Bennett, E.M.; Bohan, D.A.; Bonkowski, M.; Carpenter, S.R.; Chalmers, R.; Cramer, W.; Durance, I.;
Eisenhauer, N.; Fontaine, C.; et al. 10 Years later: Revisiting priorities for science and society a decade after
the Millennium Ecosystem Assessment. Adv. Ecol. Res. 2015, 53, 1–53. [CrossRef]
45. Bastian, M.; Heymann, S. Gephi: An open source software for exploring and manipulating networks.
In Proceedings of the Third International ICWSM Conference, San Jose, CA, USA, 17–20 May 2009;
pp. 361–362.
Sustainability 2018, 10, 4337 19 of 20

46. Barman, M.; Paul, S.; Choudhury, A.G.; Roy, P.; Sen, J. Biofertilizer as prospective input for sustainable
agriculture in India. Int. J. Curr. Microbiol. Appl. Sci. 2017, 6, 1177–1186. [CrossRef]
47. Leakey, R.R.B. The Role of Trees in Agroecology and Sustainable Agriculture in the Tropics. Annu. Rev.
Phytopathol. 2014, 1–21. [CrossRef] [PubMed]
48. Castro, A.M.; Tapias, J.; Ortiz, A.; Benavides, P.; Góngora, C.E. Identification of attractant and repellent
plants to coffee berry borer, Hypothenemus hampei. Entomol. Exp. Appl. 2017, 164, 120–130. [CrossRef]
49. Batish, D.R.; Singh, H.P.; Kohli, R.K.; Kaur, S. Eucalyptus essential oil as a natural pesticide. For. Ecol. Manag.
2008, 256, 2166–2174. [CrossRef]
50. Khadse, A. Women, Agroecology & Gender Equality; Focus on the Global South: New Delhi, India, 2017.
51. Schwendler, S.F.; Thompson, L.A. An education in gender and agroecology in Brazil’s landless rural workers’
movement. Gend. Educ. 2017, 29, 100–114. [CrossRef]
52. Altieri, M.; Toledo, V.M. The agroecological revolution in Latin America: Rescuing nature, ensuring food
sovereignty and empowering peasants. J. Peasant Stud. 2011, 38, 587–612. [CrossRef]
53. Mohri, H.; Lahoti, S.; Saito, O.; Mahalingam, A.; Gunatilleke, N.; Irham; Hoang, V.T.; Hitinayake, G.;
Takeuchi, K.; Herath, S. Assessment of ecosystem services in homegarden systems in Indonesia, Sri Lanka,
and Vietnam. Ecosyst. Serv. 2013, 5, 124–136. [CrossRef]
54. Kamiyama, C.; Hashimoto, S.; Kohsaka, R.; Saito, O. Non-market food provisioning services via homegardens
and communal sharing in satoyama socio-ecological production landscapes on Japan’s Noto peninsula.
Ecosyst. Serv. 2016, 17, 185–196. [CrossRef]
55. Fenetahun, Y.; Eshetu, G.; Worku, A.; Abdella, T. A survey on medicinal plants used by traditional healers in
Harari regional State, East Ethiopia. J. Med. Plants Stud. 2017, 5, 85–90.
56. Díaz-reviriego, A.I.; González-segura, L.; Fernández-llamazares, Á.; Howard, P.L.; Molina, J.L.;
Reyes-garcía, V. Social organization influences the exchange and species richness of medicinal plants in
Amazonian homegardens. Ecol. Soc. 2018, 21, 1. [CrossRef] [PubMed]
57. Dean, W. With Broadax and Firebrand: The Destruction of the Brazilian Atlantic Forest, 1st ed.; University of
California Press Books: Oakland, CA, USA, 1997; ISBN 9780520919082.
58. FAO. Advancing Agroforestry on the Policy Agenda; FAO: Rome, Italy, 2013.
59. Browder, J. Redemptive communities: Indigenous knowledge, colonist farming systems and conservation of
tropical forests. Agric. Hum. Values 1995, 12, 17–30. [CrossRef]
60. Perfecto, I.; Vandermeer, J. Coffee Agroecology: A New Approach to Understanding Agricultural Biodiversity,
Ecosystem Services and Sustainable Development; Routledge: New York, NY, USA, 2015; ISBN 9780415826808.
61. Khadse, A.; Rosset, P.M.; Morales, H.; Ferguson, B.G. Taking agroecology to scale: The Zero Budget Natural
Farming peasant movement in Karnataka, India. J. Peasant Stud. 2018, 45, 192–219. [CrossRef]
62. Petersen, P.F.; Silveira, L.M. Agroecology, public policies and labor-driven intensification: Alternative
development trajectories in the Brazilian semi-arid region. Sustainability 2017, 9, 535. [CrossRef]
63. Wittman, H.; Blesh, J.; Wittman, H.; Blesh, J. Food sovereignty and Fome Zero: Connecting public food
procurement programmes to sustainable rural development in Brazil. J. Agrar. Chang. 2017, 17, 81–105.
[CrossRef]
64. Miles, A.; Delonge, M.S.; Carlisle, L. Triggering a positive research and policy feedback cycle to support
a transition to agroecology and sustainable food systems. Agroecol. Sustain. Food Syst. 2017, 41, 855–879.
[CrossRef]
65. Altieri, M.; Merrick, L.C. In situ conservation of crop genetic resources through maintenance of traditional
farming systems. Econ. Bot. 1987, 41, 86–96. [CrossRef]
66. Pimbert, M. Agroecology as an alternative vision to conventional development and climate-smart Agriculture.
Development 2016, 58, 286–298. [CrossRef]
67. Rosset, P. Re-thinking agrarian reform, land and territory in La Via Campesina. J. Peasant Stud. 2013, 40,
721–775. [CrossRef]
68. Reydon, B.P.; Fernandes, V.B.; Telles, T.S. Land tenure in Brazil: The question of regulation and governance.
Land Use Policy 2015, 42, 509–516. [CrossRef]
69. Campos, A.P.T. De Conquista de Terras Em Conjunto: Redes Sociais e Confianca: A Experiencia dos
Agricultures e Agricultoras Familiares de Araponga. Master’s Thesis, Universidade Federal de Viçosa,
Viçosa, Brazil, 2006.
Sustainability 2018, 10, 4337 20 of 20

70. Pahnke, A. Institutionalizing economies of opposition: Explaining and evaluating the success of the MST’s
cooperatives and agroecological repeasantization. J. Peasant Stud. 2015, 42, 1087–1107. [CrossRef]
71. Mccune, N.; Sánchez, M. Teaching the territory: Agroecological pedagogy and popular movements.
Agric. Hum. Values 2018, 1–16. [CrossRef]
72. Pascual, U.; Balvanera, P.; Díaz, S.; Pataki, G.; Roth, E.; Stenseke, M.; Watson, R.T.; Dessane, E.B.; Islar, M.;
Kelemen, E.; et al. Valuing nature’s contributions to people: The IPBES approach. Curr. Opin. Environ. Sustain.
2017, 26, 7–16. [CrossRef]

© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).