Copyright © 2021 by the author(s). Published here under license by the Resilience Alliance.
Puri, M., E. F. Pienaar, K. K. Karanth. and B. A. Loiselle. 2021. Food for thought—examining farmers' willingness to engage in
conservation stewardship around a protected area in central India. Ecology and Society 26(2):46. https://doi.org/10.5751/
ES-12544-260246
Research
Food for thought—examining farmers' willingness to engage in conservation
stewardship around a protected area in central India
Mahi Puri 1,2, Elizabeth F. Pienaar 3,4, Krithi K. Karanth 2,5 and Bette A. Loiselle 1,6
ABSTRACT. Although protected areas (PAs) have long been considered a successful conservation strategy, more recent research has
highlighted their ecological and sociological limitations. The extant PA network is constrained by land availability and exacerbates
cultural, political, and social conflicts over access to resources. Consequently, the importance of private lands in playing a complementary
role in conservation is being widely recognized. Voluntary conservation programs that encourage private landowners to adopt
biodiversity-friendly agricultural practices have emerged worldwide. Landowners' willingness to participate in these programs is critical
to attaining landscape-level biodiversity conservation. We adopted a multidisciplinary approach, combining economic theory of rational
choice and social choice theory to explain decision making. Using a stated preference choice experiment method, we examined the role
of program design and influence of demographic, economic, and socio-psychological variables on landowners' willingness to enroll in
voluntary, incentive-based agroforestry programs. In 2018–2019, we surveyed 602 landowners in the buffer area of Pench Tiger Reserve,
India. Landowners' willingness to engage in agroforestry depended on the amount of land to be enrolled, program duration, and
incentive amount. Landowners' socio-economic characteristics, attitudes, self-efficacy, and social norms also influenced their willingness
to participate. On average, landowners required Rs. 66,000 (ca. $940 USD) per acre per year to modify their land use and adopt
agroforestry. Our study demonstrates that integrating voluntary agroforestry programs into India's rural development policy may allow
biodiversity conservation to be balanced with agricultural productivity in buffer areas surrounding PAs. We call for a new approach
that recognizes farmers as stakeholders in conservation and in creating resilient landscapes that support biodiversity and preserve
livelihoods.
Key Words: agroforestry; incentives; land sharing; private land; stated preference choice experiment; wildlife conservation
INTRODUCTION
Conservation approaches and the context for framing people–
nature relationships are evolving (Mace 2014). Although people
have long assumed that protected areas1 (PAs) are a successful
strategy for conserving the world’s biodiversity and buffering
against climate change, water insecurity, and habitat destruction
(Bruner et al. 2001), many now question the effectiveness of PAs
in achieving those goals (Mammides 2020). Although it is true
that a subset of PAs encompass biologically important ecosystems
to safeguard them from spatial and temporal threats (Andam et
al. 2008), the inadequate size, isolation, fragmentation, and
suboptimal levels of species diversity of many PAs (IUCN
categories I–II) constrain their ecological effectiveness (DeFries
et al. 2005). Studies show that the extant PA network, constrained
by the availability of land (Foley et al. 2005), is limited in its
capacity to meet conservation goals (Mora and Sale 2011). The
ranges of several species of conservation concern overlap, at least
in part, with human-dominated areas under private ownership
(Chapron et al. 2014). From a sociological perspective, researchers
have criticized the exclusionary approach used to create inviolate
PAs, which ignores the livelihoods of people who live in
surrounding areas, including their income security, access to
resources, and conflicts with wildlife (Naughton-Treves et al.
2005, West et al. 2006). To secure both biodiversity conservation
and social welfare, researchers have advocated for more voluntary
and participatory management of natural resources that includes
local people in the decision-making process (Redpath et al. 2017).
Private lands play a critical, complementary role in biodiversity
conservation at the landscape level (Kamal et al. 2015).
Accordingly, land acquisition and purchase have been regarded
as a potential approach to enhance the conservation effectiveness
of PAs by preventing land conversion to uses that do not support
conservation (McDonald-Madden et al. 2008). However, this
strategy is susceptible to criticisms of land grabbing and
displacement of communities (Ojeda 2012). Innovative
approaches that encourage voluntary conservation on private
lands have thus emerged in Europe and North America in the
form of land trusts, conservation easements, mitigation banking,
and cost-share conservation programs (Merenlender et al. 2004).
Similarly, programs such as Socio Bosque in Ecuador and Grain-
to-Green in China incentivize ecological restoration on private
lands (Chen et al. 2009, de Koning et al. 2011).
Voluntary conservation initiatives on private lands focus on
restoring the ecological integrity of agricultural or modified
landscapes. Natural regeneration is a slow process, with native
species sometimes taking decades to recolonize (Flinn and Vellend
2005). Recovery of degraded or damaged ecosystems can be
accelerated through assisted regeneration coupled with intensive
management (Chazdon 2017). Adoption of integrated land-use
systems, such as agroforestry, to create mixed landscapes is
fundamental to attaining restoration (Nair 2008). Agroforestry
combines crop production with growing trees. The trees grown
on an agroforestry landscape may support food production
(directly by providing edible products and indirectly by enhancing
1Department of Wildlife Ecology and Conservation, University of Florida, Gainesville, Florida, USA, 2Centre for Wildlife Studies, Bengaluru,
India, 3Warnell School of Forestry and Natural Resources, University of Georgia, Athens, Georgia, USA, 4Mammal Research Institute, University
of Pretoria, Pretoria, South Africa, 5Nicholas School of the Environment, Duke University, Durham, North Carolina, USA, 6Center for Latin
American Studies, University of Florida, Gainesville, Florida, USA
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
Fig. 1. A theoretical framework classifying “program factors” and “farmer (landowner)
characteristics.” The latter include land characteristics, demographic, economic, and socio-
psychological variables that can influence landowner preferences or willingness to participate in an
incentive-based conservation program. Note: this is not an exhaustive list of variables, but a
representative subset, some of which have been used in our study.
soil quality), generate materials for other economic uses (fodder,
firewood, construction), and provide medicinal materials (Nair
2008). Voluntary agroforestry initiatives that result in integrated
land-use systems align with the “land sharing” paradigm, thereby
contributing to biodiversity conservation, maintaining connectivity
across landscapes, and reducing pressures on PAs (Phalan et al.
2011, Sharma and Vetaas 2015).
Private lands conservation interventions such as agroforestry,
however, may result in opportunity costs in the form of income
foregone from alternative intensive land-use practices (Benjamin
and Sauer 2019). Individual farmers (specifically landowners)
may hamper conservation outcomes if their financial and social
needs are not met (Knight et al. 2010). Furthermore, the
effectiveness of voluntary conservation programs is contingent
on landowners’ willingness to participate in these programs
(Conradie et al. 2013). Factors that influence landowners’
willingness to engage in conservation programs include their
landholdings (e.g., property rights, land tenure, farming
practices), demographics (e.g., age, education, income), socio-
psychological characteristics (e.g., self-efficacy, trust in
government, conservation knowledge), and conservation
program design (e.g., program structure and incentives offered;
Fig. 1) (Pienaar et al. 2014, Lastra-Bravo et al. 2015, Deng et al.
2016, Lalani et al. 2016). Landowners’ engagement in
conservation programs may also be shaped by a history of people–
park or people–wildlife conflicts, institutional support afforded
to people, and the inclusiveness of the conservation decision-
making process.
The need for private lands conservation programs and the
challenges in effectively designing these programs are particularly
relevant to India. India is a megadiverse country (Mittermeier et
al. 2011) with over 800 PAs (IUCN categories II–VI) that have
succumbed to the pressures that PAs face globally. The average
size of India’s PAs is less than 200 km² (ENVIS Centre on Wildlife
and Protected Areas 2019). They are highly fragmented with
limited connectivity and are under severe development pressure.
Protected Areas in India are surrounded by high densities of
humans and livestock. Unsustainable resource use contributes to
forest degradation and negatively impacts wildlife populations
(Margulies and Karanth 2018, Li et al. 2020). Local communities
resent the State and conservation agencies because they are often
excluded from decision making, their rights over forests are
suspended, and their access to resources is limited (Kashwan
2016). As a result, alternative conservation strategies that
complement the conservation value of PAs and focus on local
involvement in decision making, including adoption of
biodiversity-friendly agriculture and ecotourism, are gaining
traction in India (Sinha et al. 2012, Ramalingam and Dharma
Rajan 2015).
Within India, a voluntary incentive-based private lands
conservation program has not been tested. Similarly, there is a
lack of science-based knowledge of the factors that facilitate or
inhibit conservation stewardship by rural communities. To
address these research gaps, we designed a study to identify how
landowners’ willingness to engage in agroforestry programs are
influenced by (1) program design, (2) the demographic and
economic characteristics of landowners, and (3) the socio-
psychological characteristics of landowners. Understanding
barriers to and motives for participation in private lands
conservation programs is necessary to improve the effectiveness
of these programs, enhance landowner participation, and
engender partnerships between local stakeholders.
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
METHODS
Study area
We conducted our study in the administrative buffer of Pench
National Park and Tiger Reserve (PTR) in Madhya Pradesh,
India (Fig. 2). The PTR covers an area of 1179.6 km² divided
between a core area (411.3 km²) and a buffer zone (768.3 km²). It
is an integral component of the central India tiger (Panthera tigris
tigris) landscape and supports tigers and a diverse assemblage of
other carnivores such as leopards (Panthera pardus), sloth bears
(Melursus ursinus), wild dogs (Cuon alpinus), herbivores including
chital (Axis axis), sambar (Rusa unicolor), nilgai (Boselaphus
tragocamelus), wild boar (Sus scrofa), and primates (Menon
2014). The core area is inviolate, with no human habitation,
resource extraction, or agriculture permitted. Regulated
economic activities (primarily agriculture) are permitted in the
buffer zone. A national highway cuts across the region.
Fig. 2. Map of the Pench Tiger Reserve (PTR), delineating the
core area and the buffer area. The farmer surveys (n = 602)
were conducted in 500 km² of buffer area to the west of the
national highway. Surveys were distributed uniformly across the
four administrative ranges and in proportion to village
population. Inset: location of PTR in the State of Madhya
Pradesh, India.
We focused our study to the west of the highway covering
approximately 500 km². There are 95 villages located within this
area, with nearly 15,000 households. The landscape is composed
of multi-use reserved forests, state-owned timber plantations, and
privately owned agricultural areas. The reserved forests are used
by people to graze livestock and collect firewood, fodder, and non-
timber forest products (NTFPs). Agriculture forms the backbone
of the local economy. The primary crops grown include wheat,
rice, maize, sugarcane, pulses, and oilseeds. Agricultural incomes
are supplemented with animal husbandry and dairy farming, sale
of NTFPs, and other service jobs provided by the government
and tourism sectors. Local communities include people belonging
to Scheduled Tribes2 (ST), Scheduled Castes (SC), Other
Backward Classes (OBC), and General category. With high
overlap between people and wildlife, negative interactions—
including livestock depredation by carnivores and crop raiding
by herbivores—are common, resulting in economic and
psychological costs.
Study design
We aimed to survey a minimum of 400 households (population
of 15,000 households) to achieve statistically significant results
(with 5% margin of error and 95% CI; Cochran 1977). We sampled
a minimum of three households in villages with less than 100
households. We proportionally increased the number of
households surveyed based on village size. We contacted
representatives from local farmers’ associations to obtain lists of
landowners (household heads) in their village. A team of two to
three enumerators (individuals who implemented the
questionnaires) administered questionnaires in-person to the
head of the household or the main land management decision
maker in the family. Every survey team included at least one local
community member to secure the trust of respondents, which was
critical to ensure research participation and honest answers. We
rigorously trained the enumerators and pre-tested the
questionnaires with 35 key informants to mitigate response biases
(e.g., strategic responses that do not reflect the true preferences
of research participants). We administered questionnaires from
November 2018 to March 2019 in the local Hindi language. We
obtained respondents’ informed oral consent before conducting
any surveys.
The questionnaire included stated preference choice experiments
(SPCEs; Hensher et al. 2005), which we designed to ascertain
respondents’ willingness to engage in agroforestry in return for
annual payments. Economists use SPCEs to elicit the value that
respondents place on different features (attributes) of a
conservation program, and the type and amount of payment
required by potential participants to enroll in the program
(Pienaar et al. 2014). Stated preference choice experiments reveal
expected levels of program participation (Adams et al. 2014) and
which characteristics of respondents influence their willingness
to participate (Harihar et al. 2015).
We designed a multi-profile SPCE, wherein respondents chose
among two profiles (or programs) with differing attribute3 levels.
Based on existing literature, we included three attributes in the
SPCEs, namely “LAND” (the percentage of land to be allocated
to agroforestry), “YEARS” (the number of years of program
enrollment), and “PAYMENT” (the amount of payment
provided per acre of land per year) (Ruto and Garrod 2009,
Scriven 2012; Table 1). Nearly 55,000 acres of land in the study
area are under private agriculture. As the percentage of land to
be enrolled by landowners is a basic unit for conservation program
implementation and because it is directly linked to landowners’
opportunity costs, we included “LAND” as our first attribute.
Agroforestry and other land restoration efforts are long-term
projects. Because such agroforestry initiatives have not been tested
in the study area before, landowners may be concerned about
enrolling in long-term, untested commitments. Accordingly, we
included “YEARS” of program enrollment as our second
attribute. The study area is drained by the Pench river and
represents a relatively fertile region, allowing for higher
agricultural incomes compared with other drier or rain-
dependent regions. As such, we included “PAYMENT” to
ascertain the monetary compensation required in our study area
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
to encourage landowner participation in conservation programs,
while still meeting the budget constraints of conservation
agencies. The levels for each attribute were determined through
key informant surveys while pre-testing the questionnaire. We
selected only three attributes to limit cognitive load for
respondents and reduce attribute non-attendance.
Table 1. Attributes and attribute levels for the stated preference
choice experiment
Attribute Attribute Levels
Percentage of land to be enrolled 25%
50%
75%
Number of years of enrollment 4 years
8 years
12 years
Payment amount per acre per year†Rs. 45,000
Rs. 60,000
Rs. 75,000
Rs. 90,000
† Rs. - Indian Rupees ($1 USD = Rs. 70 at the time of survey)
We implemented a balanced block design (Hensher et al. 2005)
to reduce respondent fatigue. With 36 possible combinations
(three land allocations x three program durations x four payment
amounts), we generated three survey blocks, each with three
choice experiments. We chose the design with the highest D-
efficiency4 of 97.19. Respondents selected one of two profiles or
the option to “opt out” by remaining at status quo, i.e., they could
choose not to enroll in any of the offered agroforestry programs
(see Fig. 3 for an example of a choice experiment). We informed
respondents that the agroforestry programs were hypothetical to
avoid false expectations of receiving any monetary benefits.
However, to avoid hypothetical bias5, we also informed
respondents that their honest responses were important to guide
the design of future conservation programs in the region. We
provided a detailed description of which land management
practices respondents would be expected to implement on
enrolled lands, specifically, cultivation of fruiting trees, medicinal
plants, and bamboo along with other crops. We used photographs
(Append. 1: Fig. A1.1) as visual aids to help respondents
understand how they would be expected to manage their land
before selecting whether they would enroll in an agroforestry
program.
In addition to the SPCEs, we collected demographic information
(e.g., age, years of education, number of household members),
economic information (e.g., size of agricultural landholdings,
income from agricultural and non-agricultural sources, diversity
in household income), and information on past interactions with
wild animals. We also collected socio-psychological data using
five-point Likert scale questions to measure respondents’
attitudes toward modifying land uses (in terms of provision of
resources, effort and expenditure required to implement
alternative land uses, and consequent conflicts with wildlife), self-
efficacy (confidence in their capabilities to meet program
requirements, availability of land, access to irrigation, technical
knowledge), and social norms (relating to family and other
community members’ support for enrolling in conservation
programs). We included the Euclidean distance between each
landholding and the PA boundary, and percentage of forest cover
(measured using the land cover map developed by Roy et al. 2015
using medium-resolution IRS LISS-III images) within a 1-km
buffer of the landholding as additional variables. All
demographic, economic, and socio-psychological variables are
described in Table A1.1 (Append. 1). We tested for collinearity in
explanatory variables prior to conducting regression analysis.
Lastly, to delve deeper into other aspects of the rural economy,
we collected ancillary data from survey respondents about rainfall
trends, emigration, forest dependence, and perceptions about
living close to a forest.
Fig. 3. An example choice experiment card used during the
surveys in Pench Tiger Reserve, Madhya Pradesh. Each
respondent was shown three such cards, with different program
options and the option to opt out (i.e., stay at status quo).
Note: Rs. – Indian Rupees ($1 USD = Rs. 70 at the time of
survey).
Analysis
We first analyzed the SPCE data using multinomial logit models
(MNL), the basic model for discrete choice modeling (Hensher
et al. 2005). Economic theory assumes that all individuals act
rationally by comparing program alternatives and choosing the
alternative that generates the greatest level of satisfaction or utility
(i.e., individuals maximize their utility). The overall utility that
individual i receives from each program (or choice profile) j (Uij)
is a function of a systematic, observable component (Vij) and a
random component (εij; McFadden 1973):
Uij =Vij +εij =X 'ij β+εij (1)
Vij =β0+βLAND∗Land ij +βYEAR ∗Yearsij
+β
P
A
Y
∗Payment
i
j
(2)
Pij =exp(X ' ijβ)
Σ
k
≠
jexp(X ' ik β)
(3)
P
ij
=exp(X'
ij
β
i
)
Σ
k
≠
j
exp(X '
ik
β
i
)
(4)
reservation price = - X 'ij β
β
P
A
Y
(5)
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
where Xij is a vector of SPCE attribute levels for program j and β
is the vector of coefficients. Accordingly, Vij took the form:
Uij =Vij +εij =X 'ij β+εij (1)
Vij =β0+βLAND∗Land ij +βYEAR ∗Yearsij
+β
P
A
Y
∗Payment
i
j
(2)
Pij =exp(X ' ijβ)
Σ
k
≠
jexp(X ' ik β)
(3)
P
ij
=exp(X'
ij
β
i
)
Σ
k
≠
j
exp(X '
ik
β
i
)
(4)
reservation price = - X 'ij β
β
P
A
Y
(5)
.
.
.
ij
ij
where “LAND” captured the percentage of land allocated to
agroforestry (25%, 50%, 75%), “YEARS” captured the duration
of enrollment in the program (4, 8, 12 years), and “PAYMENT”
captured the payment per acre per year (Rs. 45,000; Rs. 60,000;
Rs. 75,000; Rs. 90,000) presented for program j. Assuming that
the error terms follow a type I extreme value distribution, the
probability that individual i chooses program j is given by:
Uij =Vij +εij =X 'ij β+εij (1)
Vij =β0+βLAND∗Land ij +βYEAR ∗Yearsij
+β
P
A
Y
∗Payment
i
j
(2)
Pij =exp(X ' ijβ)
Σ
k
≠
jexp(X ' ik β)
(3)
P
ij
=exp(X'
ij
β
i
)
Σ
k
≠
j
exp(X '
ik
β
i
)
(4)
reservation price = - X 'ij β
β
P
A
Y
(5)
The basic multinomial logit model assumes homogeneity of
preferences across individuals, which is highly unrealistic.
Accordingly, we estimated a mixed logit (random parameters
logit; RPL) model to test for heterogeneity of preferences across
individuals, in order to better understand how respondents’
willingness to engage in agroforestry programs is influenced by
program design (research objective 1). In the RPL model, the
coefficients βi vary across individuals, but are constant across each
individual's choices (i.e., individuals are assumed to have stable
preferences; see Nordén et al. 2017, Pienaar et al. 2019):
Uij =Vij +εij =X 'ij β+εij (1)
Vij =β0+βLAND∗Land ij +βYEAR ∗Yearsij
+β
P
A
Y
∗Payment
i
j
(2)
Pij =exp(X ' ijβ)
Σ
k
≠
jexp(X ' ik β)
(3)
P
ij
=exp(X'
ij
β
i
)
Σ
k
≠
j
exp(X '
ik
β
i
)
(4)
reservation price = - X 'ij β
β
P
A
Y
(5)
The vector of random parameters βi has a mean and variance,
which captures heterogeneity of preferences across individuals. If
the standard deviation coefficient for an attribute (or attribute
level) is statistically significant, then this indicates that individuals
are heterogeneous in their preferences for that attribute (or
attribute level). We imposed a normal distribution on the β
parameters for LAND and YEARS and we assumed a fixed
parameter estimate for PAYMENT6.
To test how landowners’ willingness to engage in agroforestry
programs is influenced by their demographic and economic
characteristics and socio-psychological characteristics, we ran
two sets of RPL models to assess these drivers of preference. In
RPL model 1 (which was designed to address research objective
2), we included respondents’ demographic and economic
characteristics as shifters7 of the parameters for LAND and
YEARS to test how these characteristics drive preferences for
program design, specifically minimum land requirements to enroll
in the program and duration of enrollment. In RPL model 2
(which was designed to address research objective 3), we
interacted landowners’ socio-psychological characteristics8 with
the alternative specific constant9 (ASC), i.e., we included shifters
for β0 to test how respondents’ socio-psychological characteristics
influenced their decision to enroll in the conservation program.
We selected the best fit models based on the lowest Akaike
Information Criterion (AIC) (Burnham and Anderson 2002). We
performed all analyses in NLOGIT version 6 using a maximum
likelihood estimation procedure (Greene 2016).
Lastly, we calculated reservation payments required to enroll
landowners in the conservation programs using the fixed
parameter estimates from the best fit MNL model (Append. 1:
Table A1.4) that included respondent characteristics:
Uij =Vij +εij =X 'ij β+εij (1)
Vij =β0+βLAND∗Land ij +βYEAR ∗Yearsij
+β
P
A
Y
∗Payment
i
j
(2)
Pij =exp(X ' ijβ)
Σ
k
≠
jexp(X ' ik β)
(3)
P
ij
=exp(X'
ij
β
i
)
Σ
k
≠
j
exp(X '
ik
β
i
)
(4)
reservation price = - X 'ij β
β
P
A
Y
(5)
where βPAY is the coefficient on PAYMENT. The reservation
payment is the minimum payment required by a respondent to
engage in agroforestry. For derivation of the reservation payment,
see the appendix.
RESULTS
Demographic, economic, and socio-psychological characteristics
of households
We surveyed a total of 602 households (one person per household;
3–21 respondents per village based on the size of the village). We
administered the questionnaires to household heads, who are
often the decision makers (especially regarding land utilization).
Nearly 98% of the survey respondents were male (Append. 1:
Table A1.2). The average age of respondents was about 44 years
(range: 19–80 years), and approximately 53% of the respondents
had less than 10 years of school education. The average size of
households was five members, although we surveyed much larger
joint families (range: 1–55 household members). The average
landholding size was 11.3 acres (range: 2–90 acres). On average,
respondents grew three crops per year, consisting primarily of
food crops such as wheat, rice, and maize. Respondents could not
accurately calculate their agricultural profits because they found
it difficult to estimate total labor and input costs. Instead,
respondents provided their agricultural revenues (median= ca.
$2142 USD per year). The majority (85%) of households
supplemented their incomes from an average of two other sources
(see Append. 1: Table A1.2 for respondents’ non-agricultural
income sources and livestock ownership). See Append 1: Table
A1.3 for respondents’ perceptions of rainfall trends, forest
dependence, benefits and disadvantages of living adjacent to
forests, and emigration.
Negative interactions with wildlife can take the form of crop
raiding by herbivores and carnivores preying on livestock or
injuring and killing humans. When asked to name and rank the
three most problematic wildlife species, approximately 32% of
respondents listed only herbivores (highest conflict species: wild
pig; n = 539) and 22% of respondents also listed primates,
parakeets, snakes, and rodents (Append. 1: Table A1.2). Although
70% of respondents indicated that carnivores had injured or killed
livestock and humans in their village in the past year, only 12%
of respondents reported predator conflicts for their household
(Append. 1: Table A1.2).
The majority of respondents strongly agreed that adopting
agroforestry would provide fuelwood, fodder, and additional
income, but that agroforestry would also increase human–wildlife
conflicts (Table 2). Over half of respondents did not agree that
adopting agroforestry would increase land management costs or
require increased labor effort, and they strongly agreed that their
family would support their adoption of agroforestry (Table 2).
Respondents were most concerned that agroforestry trees would
i
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
Table 2. Percentage of responses for each rank and median responses for five-point Likert scale questions to gauge respondent’s attitudes
toward modifying land uses, self-efficacy, and social norms
Strongly disagree Moderately disagree Neutral Moderately agree Strongly agree Median †
Attitude: Do you agree or disagree that land-use change (agroforestry) would...
Provide fuelwood 10.48 0.33 2.00 16.64 70.55 5
Provide fodder 27.64 0.67 1.84 10.05 59.80 5
Provide additional income 2.66 0.67 7.32 8.32 81.03 5
Require more effort for upkeep 44.09 7.15 10.98 7.65 30.12 2
Require more expense for upkeep 54.52 5.35 10.54 6.69 22.91 1
Cause more conflict with animals 16.47 3.33 2.33 18.30 59.57 5
Self-efficacy: Do you agree or disagree that your ability to adopt agroforestry is constrained by...
Not enough land 54.58 4.16 1.50 5.66 34.11 1
No knowledge/training 29.45 4.49 1.66 9.32 55.07 5
No irrigation facility 33.11 4.66 0.33 9.65 52.25 5
Labor problems 57.02 8.03 3.51 5.85 25.59 1
Problems with neighbors 46.59 3.33 3.33 8.49 38.27 2
Trees will not grow/fruit 15.61 4.17 2.36 7.62 70.24 5
Social norms: Do you agree or disagree that the following individuals would be supportive of you adopting agroforestry...
Family 3.34 0.00 12.21 3.34 81.10 5
Other farmers 9.69 1.44 35.73 10.77 42.37 4
Forest department 14.90 1.08 37.34 4.49 42.19 3
Village council 7.79 0.56 25.79 7.05 58.81 5
†Response 1 - Strongly disagree, 2 - Moderately disagree, 3 - Neutral, 4 - Moderately agree, 5 - Strongly agree
not grow or fruit, and that they did not have the skills or necessary
irrigation to successfully implement agroforestry (Table 2).
Stated preference choice experiment
Only 12% of respondents rejected all offered agroforestry
programs. Most respondents who favored the status quo (86%)
wanted to continue their current agricultural activities and avoid
risks associated with agroforestry. Respondents who rejected
agroforestry programs also indicated that the payments were
insufficient (37%), their land was unsuitable for conservation land
uses (35%), and they were concerned about increased conflicts
with wildlife (18%).
The coefficients for all variables included in the basic MNL model
were significant at p < 0.01 or 0.05 (Table 3). On average,
respondents preferred to adopt agroforestry (positive coefficient
for the ASC). Respondents were less likely to enroll in agroforestry
as the amount of land enrolled (negative coefficient for LAND)
and the duration of enrollment (negative coefficient for YEARS)
increased. The positive coefficient for PAYMENT is consistent
with economic theory and indicates that respondents were more
likely to adopt agroforestry as the payment per acre per year for
enrollment increased.
The basic RPL model (which omitted respondent characteristics)
demonstrated that respondents were heterogeneous in their
preferences for agroforestry programs (Table 3). Respondents
preferred to engage in agroforestry relative to the status quo
option (positive mean coefficient for the ASC), although the
strength of this preference varied across respondents (statistically
significant standard deviation coefficient for the ASC). All
respondents preferred programs with smaller minimum land
requirements (negative mean coefficient for LAND), but again
the strength of this preference varied across respondents
(statistically significant standard deviation coefficient for
LAND). On average, respondents preferred shorter program
durations (negative mean coefficient for YEARS), but the
magnitude of the standard deviation coefficient demonstrated
that a subset of respondents preferred longer programs.
Respondents preferred higher payments per acre per year for
adopting agroforestry (positive coefficient for PAYMENT).
The two RPL models that allowed coefficients to shift with
respondents’ demographic, economic, and socio-psychological
variables provided greater insights into which characteristics
altered their probability of enrolling in agroforestry.
RPL model 1: interaction of demographic and economic variables
with LAND and YEARS
In this model, coefficients for LAND and YEARS shifted with
respondents’ demographic and economic characteristics and
provided further evidence that respondents preferred to enroll less
land (although the strength of this preference varied across
respondents) and that respondents differed in their preferences
for program duration (Table 4). Respondents with a high school
education and experience of crop raiding by herbivores preferred
to enroll a higher percentage of their land (p < 0.05). Respondents
with higher earnings from agriculture, small families, and higher
forest cover surrounding their agricultural land preferred to enroll
larger percentages of their land, although these results were only
significant at 0.05 < p < 0.1. Respondents with lower agricultural
earnings preferred longer programs (0.05 < p < 0.1). On average,
respondents preferred to enroll in agroforestry programs, and
their willingness to enroll increased as PAYMENT increased.
RPL model 2: interaction of socio-psychological variables with
the ASC
In this model, the ASC shifted with socio-psychological variables
and highlighted heterogeneity in respondents’ preferences for
enrolling in agroforestry programs. Although on average
respondents preferred to adopt agroforestry, a subset preferred
the status quo (standard deviation coefficient exceeded the mean
coefficient for the ASC; Table 5). Respondents who perceived that
agroforestry would generate benefits (e.g., fuelwood, fodder) were
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
Table 3. Estimated beta coefficients from multinomial logit (MNL) and random parameter logit (RPL)
analysis from choice experiment surveys conducted in the buffer area of Pench Tiger Reserve, India in 2018–
2019. The MNL model assumes homogeneity in preferences, with a single beta estimate associated with each
attribute. The RPL model assumes heterogeneity in preferences; ASC, Land and Year are considered as
normally distributed random variables, yielding a distribution of beta estimates, with a mean and variance.
Attribute MNL RPL
Mean SD
Coefficient SE Coefficient SE Coefficient SE
ASC†1.55*** 0.167 4.91*** 0.448 3.10*** 0.343
Land -0.04*** 0.002 -0.09*** 0.007 0.07*** 0.007
Years -0.02** 0.011 -0.05** 0.020 0.15*** 0.037
Payment 0.01*** 0.002 0.02*** 0.004
Model Properties
AIC/N 1.8 1.53
Log-Likelihood -1615.59 -1371.52
N (observations) 1800 1800
†ASC is an alternative specific constant taking the value 1 if one of the agroforestry programs (Program A or Program B) is
chosen and zero otherwise (Program C)
*** p < 0.01; ** p < 0.05; * 0.05 < p < 0.1
more likely to participate (p < 0.01). Respondents who did not
view their landholding size as a constraint to agroforestry, and
those who were concerned about access to irrigation were also
more likely to participate (p < 0.05 and 0.05 < p < 0.1,
respectively). Perceived support from family members increased
the likelihood that respondents would choose to adopt
agroforestry (0.05 < p < 0.1). Respondents preferred to enroll
fewer acres for a shorter duration and preferred a higher annual
payment per acre for adopting agroforestry.
We used the best fit MNL model (Append. 1: Table A1.4) to
calculate the minimum annual payment per acre (reservation
payment) respondents required to enroll in an agroforestry
program. The average reservation payment was approximately Rs.
66,000/acre/year (ca. $940 USD/acre/year; $1 USD = Rs. 70 at
the time of survey).
DISCUSSION
The central Indian landscape supports one of the world’s largest
tiger populations and is recognized as a global priority landscape
for tigers (Wikramanayake et al. 2011). It comprises a network of
16 PAs, some connected by remnant or degraded forests. Apart
from the critical role played by PAs in the region, multi-use
reserved forests, scrublands, and degraded lands are also highly
used by a variety of mammals (Dutta et al. 2015, Srivathsa et al.
2019, Puri et al. 2020). Despite the creation of PAs, the central
India landscape has the highest levels of fragmentation due to
linear intrusions (roads, railways, and powerlines) with more
isolated forest patches relative to other regions in the country
(Nayak et al. 2020). A large percentage of the landscape, crucial
for maintaining connectivity, is not permeable to animal
movement as it is restricted by human land-use, human
population, and high density of linear infrastructure (Jayadevan
et al. 2020). Even “impediment-free” areas are fragmented and
surrounded by areas that inhibit animal movement, e.g., low cover
agricultural areas. Several studies focusing on central India
recommend maintaining and improving the connectivity of the
landscape through restoration of the habitat (Rathore et al. 2012,
Joshi et al. 2013, Yumnam et al. 2014, Dutta et al. 2018). As a
significant proportion of the land outside PAs is under private
ownership and used for agriculture, long-term persistence of
biodiversity in the central Indian landscape is contingent on the
way this agricultural matrix is managed, with proper incentives
and technical advice.
Restoration of degraded lands has seen some success in India,
albeit on communally managed lands with the recognition of
community forest resource (CFR) rights under the Forest Rights
Act, 2006. Communities have experimented with short-rotation
species (bamboo and various fruiting trees), generated incomes
through harvest of NTFP, initiated measures for soil and water
conservation, imposed restrictions on grazing activities in areas
under assisted natural regeneration, and even set aside areas to
allow wildlife presence (Agarwal and Saxena 2018). There are few
examples of private or NGO-led conservation initiatives, at small
spatial scales (Mudappa and Raman 2007). In production
landscapes, sustaining biodiversity with agricultural productivity
is challenging. Integration of private agricultural lands—a
neglected constituency—into India’s conservation framework can
go a long way in reconciling agriculture, biodiversity, and rural
livelihoods (Siebert et al. 2006, Chen et al. 2009, Scriven 2012).
To help inform these conservation efforts, we designed a study in
which we combined economic models of rational choice (i.e.,
utility maximization) with broader concepts from social choice
theory to examine landowners’ willingness to adopt agroforestry
in the buffer area of Pench National Park and Tiger Reserve. We
focused on how program design and demographic, economic, and
socio-psychological variables influenced landowners’ willingness
to voluntarily enter agroforestry programs.
Program design
Landowner participation in voluntary conservation programs
depends on program design (Ruto and Garrod 2009, Espinosa-
Goded et al. 2010), in particular whether programs are relevant
to local contexts. We found that percentage of land enrolled,
duration of enrollment, and payment amount significantly
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
Table 4. Estimated beta coefficients for a random parameter logit
(RPL) with interaction variables. We allowed Land and Year to
be normally distributed random variables. We also included fixed
parameters in the model that interacted Land and Year with the
demographic and economic characteristics of respondents from
choice experiment surveys conducted in the buffer area of Pench
Tiger Reserve, India in 2018–2019. The sign of the coefficients
indicates the correlation between program attributes and
landowner characteristics. Estimates are from the best-fit model
based on AIC.
Mean SD
Coefficient SE Coefficient SE
ASC†4.43*** 0.343
Land -0.16*** 0.03 0.07*** 0.006
Years 0.09 0.127 0.23*** 0.03
Payment 0.02*** 0.004
Land X FCOV 0.03* 0.017
Land X CONF 0.05** 0.018
Land X AGE 0.004 0.009
Land X EDU1 0.01 0.012
Land X EDU2 -0.01 0.016
Land X EDU3 0.04** 0.017
Land X EDU4 0.005 0.017
Land X HHS -0.002* 0.001
Land X SIZE 0.0001 0.002
Land X AGIN 0.0003* 2E-04
Land X CROP -0.003 0.004
Land X SOUR -0.001 0.004
Years X FCOV -0.04 0.078
Years X CONF -0.07 0.076
Years X AGE -0.04 0.039
Years X EDU1 -0.07 0.054
Years X EDU2 -0.02 0.071
Years X EDU3 -0.09 0.081
Years X EDU4 -0.11 0.075
Years X HHS 0.002 0.006
Years X SIZE 0.01 0.008
Years X AGIN -0.002* 9E-04
Years X CROP 0.01 0.02
Years X SOUR 0.006 0.018
Model Properties
AIC/N 1.568
Log-Likelihood -1376.94
FCOV - percentage forest cover in 1 km of landholding, CONF - dummy
variable takes a value of 1 if respondent has experienced conflict with
herbivore, AGE - dummy variable takes a value of 1 if respondent is
more than 44 years (average age), EDU1 - Education of respondent less
than 10th grade, EDU2 - Passed 10th grade, EDU3 - Passed 12th grade,
EDU4 - Graduate and above, HHS - Household size, SIZE - Size of
landholding, AGIN - Income from agriculture, CROP - Number of
crops grown, SOUR - Number of income sources
†ASC is an alternative specific constant taking the value 1 if one of the
agroforestry programs (Program A or Program B) is chosen and zero
otherwise (Program C)
*** p < 0.01; ** p < 0.05; * 0.05 < p < 0.1
influenced landowners’ willingness to adopt agroforestry. Our
finding that respondents preferred to enroll smaller percentages
of their land for shorter contract durations (see also Ruto and
Garrod 2009) was consistent with comments by several
landowners during field research that they would prefer to choose
the least restrictive program design (25% land enrolled for 4 years)
and if the program proved beneficial (e.g., consistent payments,
Table 5. Estimated beta coefficients for a random parameter logit
(RPL) with interactions. We allowed the alternative specific
constant (ASC) to be a normally distributed random variable. We
also included fixed parameters in the model that interacted the
ASC with the socio-psychological characteristics of respondents
from choice experiment surveys conducted in the buffer area of
Pench Tiger Reserve, India in 2018–2019. The sign of the
coefficients indicates the correlation between landowner
characteristics and their preferences. Estimates are from the best-
fit model based on AIC.
Mean SD
Coefficient SE Coefficient SE
ASC†1.71*** 0.291 2.06*** 0.175
Land -0.05*** 0.002
Years -0.02* 0.013
Payment 0.02*** 0.003
ASC X LINP -0.005 0.123
ASC X LBNF 0.66*** 0.124
ASC X LSIZ -0.30*** 0.068
ASC X LKNW -0.002 0.067
ASC X LLBR -0.11 0.071
ASC X LIRR 0.12* 0.067
ASC X LFAM 0.25* 0.129
Model Properties
AIC/N 1.615
Log Likelihood -1407.84
LINP - Component generated using PCA combining Likert statements
relating to additional requirements of effort and financial expenditure,
LBNF - Component generated using PCA combining Likert statements
relating to provision of fodder, firewood, and additional income, LSIZ -
Likert response related to self-efficacy/worry about land size, LKNW -
Likert response related to self-efficacy/worry about technical knowledge/
training, LLBR - Likert response related to self-efficacy/worry about
labor availability, LIRR - Likert response related to self-efficacy/worry
about access to irrigation, LFAM - Likert response related to social
norms/ agreement with family members
†ASC is an alternative specific constant taking the value 1 if one of the
agroforestry programs (Program A or Program B) is chosen and zero
otherwise (Program C)
*** p < 0.01; ** p < 0.05; * 0.05 < p < 0.1
increased access to resources) then they would enroll more land
for longer durations in future. Landowners may be skeptical about
enrolling land for a long duration due to incomplete knowledge
of future benefits, lack of trust that the program will be funded
in the medium or long term, and associated risks (e.g., failure of
trees to grow and fruit) and costs (e.g., reduced capacity to
produce food crops) with transitioning to agroforestry. As there
is no existing proof of concept for a voluntary agroforestry
program in India, conservation contract designs should allow
flexibility in the amount of land enrolled and enrollment duration,
provided that the conservation impact of the program is not
fundamentally compromised.
Consistent with economic theory, respondents preferred
programs with higher payments. Landowners may view
conservation payments as a steady, alternative source of income
that allows them to reduce risks associated with land-use change,
pay household expenses, or invest in off-farm activities (Siebert
et al. 2006, Jones et al. 2017). During surveys, respondents
indicated that they would invest conservation payments in
improved irrigation systems and fencing. Some landowners
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
suggested that rather than receiving fixed annual payments, they
would prefer larger payments in the initial years to make necessary
land-use changes and reduced payments later. Payment structure,
frequency of payments, and terms for defaulting and cancellation
are important considerations when designing an effective
conservation program.
Demographic and economic determinants of program
participation
High surrounding forest cover (which would likely be associated
with increased wildlife density) and crop raiding by herbivores
increased respondents’ willingness to adopt agroforestry. In India,
landowners are under-compensated for losses due to crop raiding
and often absorb the financial losses themselves (Karanth et al.
2018). A program that provides incentives to adopt agroforestry
practices that are not susceptible to crop raiding by herbivores
was therefore financially attractive to respondents. Diversification
to a multi-crop agroforestry system would also reduce the risks
of single crop agriculture. However, adoption of agroforestry
systems may result in higher levels of human–wildlife interactions
(e.g., livestock predation) within villages. A successful
collaborative conservation program will require timely aid to
farmers when they face losses from wildlife. Negotiating
thresholds or acceptable limits to losses will be challenging but
necessary for successful implementation of agroforestry
programs.
Landowners with higher education levels are more easily able to
engage in the necessary training and process paperwork
associated with voluntary conservation programs (Siebert et al.
2006, Peerlings and Polman 2009). We found high school
education was the threshold level that encouraged farmers to
enroll more land in agroforestry. By contrast, respondents with
larger families preferred to enroll less land, which is likely because
large rural families in India require more land to meet their
subsistence needs and land must be divided amongst multiple
male successors (see also Lastra-Bravo et al. 2015 on the role of
social networks and successors in the decision to adopt
conservation practices). Landowners with higher agricultural
earnings preferred to enroll more land in agroforestry perhaps
because they could absorb opportunity costs. However,
landowners with lower agricultural earnings preferred to enroll
for more years, likely to secure payments over a longer period that
would allow them to meet their household’s financial needs. In
contrast to Scriven (2012), we did not find evidence that
respondents with larger landholdings were willing to enroll larger
portions of their land. Our results also did not support Broch and
Vedel’s (2012) finding that farmers who rely only on agricultural
incomes require higher compensation to enroll in conservation
programs.
Socio-psychological determinants of program participation
Financial incentives may not secure long-term enrollment in
conservation programs because they are only an external
motivator for voluntary behavioral change (Siebert et al. 2006).
We found that payments were not sufficient to persuade
landowners who want to remain financially independent by
continuing their existing agricultural practices to participate in
agroforestry (also see Schenk et al. 2007). The 12% of respondents
who rejected all agroforestry programs were primarily concerned
about their ability to meet their household’s subsistence needs and
ensure their food security. Financial incentives were insufficient
to overcome these concerns. Respondents also feared land-use
change would result in interference, control, or illegal acquisition
of private land by the forest department, which is consistent with
previous findings that lack of trust in government impedes
voluntary participation in conservation programs (Scriven et al.
2012, Jones et al. 2017).
Inadequate information about the benefits of conservation
programs is a barrier to voluntary landowner enrollment in these
programs (Kabii and Horwitz 2006). We found that perceived
benefits from agroforestry (such as increased availability of
fuelwood, fodder, and income), greater self-efficacy, and perceived
support of family members for enrollment in conservation
programs (social norms) increased respondents’ willingness to
participate in agroforestry. Self-efficacy, reinforced by availability
of resources, training, and social capital, may increase voluntary
adoption of agroforestry (McGinty et al. 2008). However,
respondents expressed concerns that they had insufficient land to
engage in an agroforestry program and lacked the necessary
technical knowledge or training to successfully adopt
agroforestry. Agricultural systems across India are typically
managed by smallholders. Altough this may seem a challenge to
reconciling scale-dependent environmental benefits, studies have
shown that small farms often have a high capacity to sustain
biodiversity and rural livelihoods (Kumaraswamy and Kunte
2013). Interestingly, respondent concerns that agroforestry would
fail on their properties because they lack irrigation increased their
willingness to adopt agroforestry. Although this initially seems
counter-intuitive, some landowners viewed conservation
payments as an opportunity to invest in infrastructure (such as
wells) on their land. Our results suggest that capacity building is
important for ensuring the success of agroforestry development
programs. Outreach activities focusing on knowledge sharing and
skill development may be crucial in shaping landowner attitudes,
alleviating concerns about the viability of conservation actions
and improving community-level acceptance of programs (van den
Berg et al. 2011, Ardoin et al. 2020).
CONCLUSION
Over the last five decades, India has conserved its biodiversity
through the creation of state-controlled PAs. India’s conservation
policy has largely ignored the need for a complementary strategy
of land sharing in human-dominated landscapes (Fischer et al.
2014, Robbins et al. 2015). Integrated land-use systems would
allow conservation and production units to be co-managed for
long-term sustainability and improved social welfare (Harvey et
al. 2008). However, restoring degraded agricultural landscapes is
expensive and typically relies on a mix of government funds,
private investments, and NGO support (Clarvis 2014).
Well-designed payments for ecosystem services (PES) may
voluntarily engage landowners in land-use practices such as
agroforestry that secure biodiversity and ecosystem services
(Montagnini and Finney 2011). Within the context of the ongoing
agrarian crisis in India, Devi et al. (2017) have argued that PES
may also bring farmers out of poverty traps by providing a fixed
income. They estimated that the total economic value of the
ecosystem services provided by cultivated agroecosystems is
approximately Rs. 71,000 (ca. $1015 USD) per acre per year. We
found that farmers in the buffer areas of Pench required an
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
average of Rs. 66,000 (ca. $940 USD) per acre per year to adopt
agroforestry, almost equivalant to Devi et al.’s (2017) estimates
of conservation benefits that would be generated from
agroforestry. Although scholars have critiqued PES approaches
(Lele et al. 2010) and have questioned whether India can support
the institutional framework necessary for PES implementation
(Sharma 2017), there is a consensus that increased farm incomes
are urgently needed. Our study demonstrates the potential for an
agroforestry-based PES program to be integrated into the
country’s agroforestry and rural development policy (Siebert et
al. 2006, de Koning et al. 2011). However, we do caution that the
effectiveness of an agroforestry PES program will likely depend
on flexibility of program design and capacity building. Socio-
economic constraints may preclude the enrollment of individuals
who do not have the financial means to alter their land
management practices, which means that landowners who are
most at risk because of poverty may be least likely to engage in
an agroforestry PES program. Payment structure may need to
vary based on households’ financial and resource constraints.
Although it is not immediately apparent how an agroforestry PES
program would be financed, our study suggests that landowners
would be interested in adopting agroforestry if the program is
appropriately designed for social and economic contexts. We
envision the application of incentive-based land management
practices in fragmented landscapes to restore connectivity and
increase the effective size of PAs by improving the ecological
quality of surrounding lands. Kshettry et al. (2020) proposed the
term “Conservation Compatible Landscape” (CCL) to denote
landscapes with high potential for conservation in conjunction
with local support. An approach such as ours, which focuses on
private landowner willingness to become conservation stewards,
can be a means toward realizing these CCLs. We call for a new
approach in India that recognizes farmers as stakeholders in
conservation and in creating resilient landscapes that support
biodiversity and preserve rural livelihoods.
---------------
1 As described by the International Union for Conservation of
Nature (IUCN), Protected Areas are “a clearly defined
geographical space, recognized, dedicated, and managed, through
legal or other effective means, to achieve the long-term
conservation of nature with associated ecosystem services and
cultural values.” They are categorized based on their management
objectives, with categories I and II representing areas where
human visitation, use, and impacts are strictly controlled and
limited.
2 Tribal communities and other caste groups (including SC and
OBC) are part of the hierarchical caste system of India. These
communities are listed in a schedule prepared by the Government
of India, granting them special status by the Constitution.
Reservations in legislature and government jobs are based on this
categorization (Bhargava 2010).
3 A profile (or program) is an alternative that is offered to a person.
The program features are called “attributes,” and these attributes
vary across profiles. For example, a SPCE that focuses on
respondents’ choice of travel may vary according to the attributes
of travel mode (car, train, and bus), travel time, and cost (Hensher
et al. 2005).
4 D-efficiency is related to the statistical efficiency of the design
or the attribute combinations that are generated. The best design
is the one with the highest D-efficiency, out of a score of 100 (for
more details, see Append. 1).
5 Hypothetical bias occurs when respondents provide hypothetical
answers (something they have no intention of doing or are not
able to do owing to constraints on their behavior, e.g., budget or
labor constraints) when presented with a SPCE.
6 We note that it is possible to relax the assumption of a fixed
parameter on the payment attribute by imposing a lognormal
distribution for the coefficient on PAYMENT. However, we found
that the random parameters logit model that assumed a fixed
parameter for PAYMENT was comparable in terms of model fit
to that in which the coefficient on PAYMENT varied. Moreover,
the model with a random coefficient for PAYMENT generated
inflated measures of respondents’ willingness to accept
compensation for enrollment in an agroforestry program.
Accordingly, we did not relax the assumption of a fixed parameter
for PAYMENT.
7 We included respondents’ characteristics in the estimated models
as interaction terms (e.g., respondents’ characteristics are
interacted with program attributes to test for changes in the slope
of the estimated function) or as stand-alone independent
variables (to test for changes in the intercept of the estimated
function).
8 To reduce dimensionality, we used principal component analysis
with varimax rotation to ascertain if individual survey items could
be combined to form measures (scores) of respondents’ attitudes
toward agroforestry. We retained two components with
eigenvalues≥1 and Cronbach’s alpha≥0.7. The two components
represented (a) perceived benefits of agroforestry from increased
availability of firewood and fodder and increased income levels,
and (b) perceived additional expenditures and effort for upkeep/
maintenance required to successfully engage in agroforestry. The
scores for the two components were calculated and used as
predictor variables in the SPCE models. Based on principal
components analysis and Cronbach’s alpha, statements that were
designed to measure self-efficacy and social norms could not be
combined into scores. Accordingly, we included each of these
individual items in the SPCE models after testing for collinearity.
9 For the models we estimated, the alternative specific constant
(ASC) took the value 1 if one of the choice scenarios was chosen
and zero otherwise. As such, if β0 (the estimated coefficient for
the ASC) is positive, then respondents preferred an agroforestry
program over the status quo of no conservation program.
Responses to this article can be read online at:
https://www.ecologyandsociety.org/issues/responses.
php/12544
Author Contributions:
Mahi Puri: conceptualization, funding acquisition, methodology,
formal analysis, investigation, visualization, writing - original draft,
review, and editing; Elizabeth Pienaar: funding acquisition,
methodology, validation, writing - review, and editing, supervision;
Krithi Karanth: funding acquisition, writing - review, and editing,
supervision; Bette Loiselle: funding acquisition, writing - review,
and editing, supervision.
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
Acknowledgments:
We thank the Madhya Pradesh Forest Department for providing
the necessary research permits to conduct this study. We received
funding from National Geographic Society (early career grant),
the Rufford Foundation, and DeFries-Bajpai Foundation. The
funding agencies had no role in study design, in the collection,
analysis, and interpretation of data, and in the decision to submit
the article for publication. MP was supported by the University of
Florida and KK was supported by Oracle. Centre for Wildlife Studies
provided institutional and logistical support. We are grateful to A.
Adambey, A. Bhatia, A. S. Chauhan, A. Pattnaik, A. Shaikh, C.
Hinge, F. Mookherjee, G. Vijayraghavan, N. Bomcher, P. Thakre,
V. Sabharwal, Y. Khatri for assistance in conducting field surveys.
Data Availability:
The data that support the findings of this study are available on
request from the corresponding author. None of the data are publicly
available as they contain information that could compromise the
privacy of research participants.
LITERATURE CITED
Adams, V. M., R. L. Pressey, and N. Stoeckl. 2014. Estimating
landholders’ probability of participating in a stewardship
program, and the implications for spatial conservation priorities.
PLoS One 9(6): e97941. https://doi.org/10.1371/journal.
pone.0097941
Agarwal, S., and A. K. Saxena. 2018. People’s forests: is
community forest resource governance the future of India’s
jungles? Centre for Science and Environment, New Delhi, India.
Andam, K. S., P. J. Ferraro, A. Pfaff, G. A. Sanchez-Azofeifa, and
J. A. Robalino. 2008. Measuring the effectiveness of protected
area networks in reducing deforestation. Proceedings of the
National Academy of Sciences 105(42):16089-16094. https://doi.
org/10.1073/pnas.0800437105
Ardoin, N. M., A. W. Bowers, and E. Gaillard. 2020.
Environmental education outcomes for conservation: a
systematic review. Biological Conservation 241: 108224. https://
doi.org/10.1016/j.biocon.2019.108224
Benjamin, E. O., and J. Sauer. 2018. The cost effectiveness of
payments for ecosystem services—smallholders and agroforestry
in Africa. Land Use Policy 71:293-302. https://doi.org/10.1016/j.
landusepol.2017.12.001
Bhargava, R. 2010. The promise of India's secular democracy.
Oxford University Press, Oxford, UK.
Broch, S. W., and S. E. Vedel. 2012. Using choice experiments to
investigate the policy relevance of heterogeneity in farmer agri-
environmental contract preferences. Environmental and
Resource Economics 51(4):561-581. https://doi.org/10.1007/
s10640-011-9512-8
Bruner, A. G., R. E. Gullison, R. E. Rice, and G. A. B. Fonseca.
2001. Effectiveness of parks in protecting tropical biodiversity.
Science 291:125-128. https://doi.org/10.1126/science.291.5501.125
Burnham, K. P., and D. R. Anderson. 2002. Model selection and
multimodel inference: a practical information-theoretic
approach. Springer, New York, New York, USA.
Chapron, G., P. Kaczensky, J. D. C. Linnell, M. von Arx, D. Huber,
H. Andrén, J. V. López Bao, M. Adamec, F. Álvares, O. Anders
et al. 2014. Recovery of large carnivores in Europe’s modern
human-dominated landscapes. Science 346(6216):1517-1519.
Chazdon, R. L. 2017. Landscape restoration, natural
regeneration, and the forests of the future. Annals of the Missouri
Botanical Garden 102(2):251-257. https://doi.org/10.3417/2016035
Chen, X., F. Lupi, G. He, and J. Liu. 2009. Linking social norms
to efficient conservation investment in payments for ecosystem
services. Proceedings of the National Academy of Sciences 106
(28):11812-11817. https://doi.org/10.1073/pnas.0809980106
Clarvis, M. H. 2014. Institutions and mechanisms that finance
ILM. Pages 19-26 in S. Shames, M. H. Clarvis, and G. Kissinger,
editors. Financing strategies for integrated landscape investment,
EcoAgriculture Partners, on behalf of the Landscapes for People,
Food and Nature Initiative, Washington, D.C., USA.
Cochran, W. G. 1977. Sampling techniques. Third edition. Wiley,
New York, New York, USA.
Conradie, B., M. Treurnicht, K. Esler, and M. Gaertner. 2013.
Conservation begins after breakfast: the relative importance of
opportunity cost and identity in shaping private landholder
participation in conservation. Biological Conservation 158:334-341.
https://doi.org/10.1016/j.biocon.2012.08.028
DeFries, R., A. Hansen, A. C. Newton, and M. C. Hansen. 2005.
Increasing isolation of protected areas in tropical forests over the
past twenty years. Ecological Applications 15(1):19-26. https://
doi.org/10.1890/03-5258
de Koning, F., M. Aguiñaga, M. Bravo, M. Chiu, M. Lascano, T.
Lozada, and L. Suarez. 2011. Bridging the gap between forest
conservation and poverty alleviation: the Ecuadorian Socio
Bosque program. Environmental Science and Policy 14:531-542.
https://doi.org/10.1016/j.envsci.2011.04.007
Deng, J., P. Sun, F. Zhao, X. Han, G. Yang, and Y. Feng. 2016.
Analysis of the ecological conservation behavior of farmers in
payment for ecosystem service programs in eco-environmentally
fragile areas using social psychology models. Science of the Total
Environment 550:382-390. https://doi.org/10.1016/j.scitotenv.2016.01.152
Devi, I. P., L. Kumar L., D. S. Kumar, M. Manjula, P.
Mukhopadhyay, D. Sharma, R. Sridhar, and L. Venkatachalam.
2017. Payment for ecosystem services: guaranteed farm income
and sustainable agriculture. Economic and Political Weekly 52
(17):12-14.
Dutta, T., S. Sharma, and R. DeFries. 2018. Targeting restoration
sites to improve connectivity in a tiger conservation landscape in
India. PeerJ 6: e5587. https://doi.org/10.7717/peerj.5587
Dutta, T., S. Sharma, J. E. Maldonado, H. S. Panwar, and J.
Seidensticker. 2015. Genetic variation, structure, and gene flow
in a sloth bear (Melursus ursinus) meta-population in the Satpura-
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
Maikal landscape of central India. PLoS One 10(5): e0123384.
https://doi.org/10.1371/journal.pone.0123384
ENVIS Centre on Wildlife and Protected Areas. 2019. Protected
areas of India. Wildlife Institute of India, Dehradun, India.
[online] URL: http://www.wiienvis.nic.in/Database/Protected_Area_854.
aspx
Espinosa-Goded, M., J. Barreiro-Hurlé, and E. Ruto. 2010. What
do farmers want from agri-environmental scheme design? A
choice experiment approach. Journal of Agricultural Economics
61(2):259-273. https://doi.org/10.1111/j.1477-9552.2010.00244.x
Fischer, J., D. J. Abson, V. Butsic, M. J. Chappell, J. Ekroos, J.
Hanspach, T. Kuemmerle, H. G. Smith, and H. von Wehrden.
2014. Land sparing versus land sharing: moving forward.
Conservation Letters 7(3):149-157. https://doi.org/10.1111/
conl.12084
Flinn, K. M., and M. Vellend. 2005. Recovery of forest plant
communities in post‐agricultural landscapes. Frontiers in
Ecology and the Environment 3(5):243-250. https://doi.
org/10.1890/1540-9295(2005)003[0243:rofpci]2.0.co;2
Foley, J. A., R. DeFries, G. P. Asner, C. Barford, G. Bonan, S. R.
Carpenter, F. S. Chapin, M. T. Coe, G. C. Daily, H. K. Gibbs, et
al. 2005. Global consequences of land use. Science 309
(5734):570-574. https://doi.org/10.1126/science.1111772
Green, W. H. 2016. NLOGIT version 6. Econometric Software
Inc., New York, New York, USA.
Harihar, A., D. Veríssimo, and D. C. MacMillan. 2015. Beyond
compensation: integrating local communities’ livelihood choices
in large carnivore conservation. Global Environmental Change
33:122-130. https://doi.org/10.1016/j.gloenvcha.2015.05.004
Harvey, C. A., O. Komar, R. Chazdon, B. G. Ferguson, J. B.
Finegan, D. M. Griffith, M. Martínez-Ramos, H. Morales, R.
Nigh, L. Soto-Pinto et al. 2008. Integrating agricultural
landscapes with biodiversity conservation in the Mesoamerican
hotspot. Conservation Biology 22(1):8-15. https://doi.org/10.1111/
j.1523-1739.2007.00863.x
Hensher, D. A., Green, W. H., and J. M. Rose. 2005. Applied
choice analysis: a primer. Cambridge University Press. New York,
New York, USA. https://doi.org/10.1017/CBO9780511610356
Jayadevan, A., R. Nayak, K. K. Karanth, J. Krishnaswamy, R.
DeFries, K. U. Karanth, and S. Vaidyanathan. 2020. Navigating
paved paradise: Evaluating landscape permeability to movement
for large mammals in two conservation priority landscapes in
India. Biological Conservation 247:108613. https://doi.org/10.1016/
j.biocon.2020.108613
Jones, K. W., M. B. Holland, L. Naughton-Treves, M. Morales,
L. Suarez, and K. Keenan. 2017. Forest conservation incentives
and deforestation in the Ecuadorian Amazon. Environmental
Conservation 44(1):56-65. https://doi.org/10.1017/S0376892916000308
Joshi, A., S. Vaidyanathan, S. Mondol, A. Edgaonkar, and U.
Ramakrishnan. 2013. Connectivity of tiger (Panthera tigris)
populations in the human-influenced forest mosaic of Central
India. PloS One 8(11): e77980. https://doi.org/10.1371/journal.
pone.0077980
Kabii, T., and P. Horwitz. 2006. A review of landholder
motivations and determinants for participation in conservation
covenanting programmes. Environmental Conservation 33:11-20.
https://doi.org/10.1017/S0376892906002761
Kamal, S., M. Grodzińska-Jurczak, and G. Brown. 2015.
Conservation on private land: a review of global strategies with
a proposed classification system. Journal of Environmental
Planning and Management 58(4):576-597. https://doi.
org/10.1080/09640568.2013.875463
Karanth, K. K., S. Gupta, and A. Vanamamalai. 2018.
Compensation payments, procedures and policies towards
human-wildlife conflict management: Insights from India.
Biological Conservation 227:383-389. https://doi.org/10.1016/j.
biocon.2018.07.006
Kashwan, P. 2016. Power asymmetries and institutions: landscape
conservation in central India. Regional Environmental Change
16(S1):97-109. https://doi.org/10.1007/s10113-015-0925-8
Knight, A. T., R. M. Cowling, M. Difford, and B. M. Campbell.
2010. Mapping human and social dimensions of conservation
opportunity for the scheduling of conservation action on private
land. Conservation Biology 24: 1348-58. https://doi.org/10.1111/
j.1523-1739.2010.01494.x
Kshettry, A., S. Vaidyanathan, R. Sukumar, and V. Athreya. 2020.
Looking beyond protected areas: identifying conservation
compatible landscapes in agro-forest mosaics in north-eastern
India. Global Ecology and Conservation 22: e00905. https://doi.
org/10.1016/j.gecco.2020.e00905
Kumaraswamy, S., and K. Kunte. 2013. Integrating biodiversity
and conservation with modern agricultural landscapes.
Biodiversity and Conservation 22:2735-2750. https://doi.
org/10.1007/s10531-013-0562-9
Lalani, B., P. Dorward, G. Holloway, and E. Wauters. 2016.
Smallholder farmers' motivations for using conservation
agriculture and the roles of yield, labour and soil fertility in
decision making. Agricultural Systems 146:80-90. https://doi.
org/10.1016/j.agsy.2016.04.002
Lastra-Bravo, X.B., C. Hubbard, C., G. Garrod, and A. Tolón-
Becerra. 2015. What drives farmers’ participation in EU agri-
environmental schemes?: results from a qualitative meta-analysis.
Environmental Science and Policy 54:1-9. https://doi.
org/10.1016/j.envsci.2015.06.002
Lele, S., P. Wilshusen, D. Brockington, R. Seidler, and K. Bawa.
2010. Beyond exclusion: alternative approaches to biodiversity
conservation in the developing tropics. Current Opinion in
Environmental Sustainability 2(1-2):94-100. https://doi.org/10.1016/
j.cosust.2010.03.006
Li, B. V., K. Reardon, N. Satheesh, C. Liu, and K. K. Karanth.
2020. Effects of livestock loss and emerging livestock types on
livelihood decisions around protected areas: case studies from
China and India. Biological Conservation 248: 108645. https://
doi.org/10.1016/j.biocon.2020.108645
Mace, G. M. 2014. Whose conservation? Changes in the
perception and goals of nature conservation require a solid
scientific basis. Science 245(6204):1558-1560.
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
Mammides, C. 2020. A global analysis of the drivers of human
pressure within protected areas at the national level. Sustainability
Science 15:1223-1232. https://doi.org/10.1007/s11625-020-00809-7
Margulies, J. D., and K. K. Karanth. 2018. The production of
human-wildlife conflict: a political animal geography of
encounter. Geoforum 95:153-164. https://doi.org/10.1016/j.
geoforum.2018.06.011
McDonald-Madden, E., M. Bode, E. T. Game, H. Grantham,
and H. P. Possingham. 2008. The need for speed: informed land
acquisitions for conservation in a dynamic property market.
Ecology Letters 11(11):1169-1177. https://doi.org/10.1111/
j.1461-0248.2008.01226.x
McFadden, D. 1973. Conditional logit analysis of qualitative
choice behavior. Pages 105-142 in P. Zarembka, editor. Frontiers
in econometrics. Academic Press, New York, New York, USA.
McGinty, M. M., M. E. Swisher, and J. Alavalapati. 2008.
Agroforestry adoption and maintenance: self-efficacy, attitudes
and socio-economic factors. Agroforestry Systems 73:99-108.
https://doi.org/10.1007/s10457-008-9114-9
Menon, V. 2014. Indian mammals: a field guide. Hatchett India,
Gurgaon, India.
Merenlender, A.M., L. Huntsinger, G. Guthey, G., and S. K.
Fairfax. 2004. Land trusts and conservation easements: who is
conserving what for whom? Conservation Biology 18(1):65-76
https://doi.org/10.1111/j.1523-1739.2004.00401.x
Mittermeier, R. A., W. R. Turner, F. W. Larsen, T. M. Brooks,
and C. Gascon. 2011. Global biodiversity conservation: the
critical role of hotspots. Pages 3-22 in F. Zachos and J. Habel,
editors. Biodiversity hotspots. Springer, Heidelberg, Germany.
https://doi.org/10.1007/978-3-642-20992-5_1
Montagnini, F., and C. Finney. 2011. Payments for environmental
services in Latin America as a tool for restoration and rural
development. Ambio 40(3):285-297. https://doi.org/10.1007/
s13280-010-0114-4
Mora, C., and P. F. Sale. 2011. Ongoing global biodiversity loss
and the need to move beyond protected areas: a review of the
technical and practical shortcomings of protected areas on land
and sea. Marine Ecology Progress Series 434:251-266. https://doi.
org/10.3354/meps09214
Mudappa, D., and T. S. Raman. 2007. Rainforest restoration and
wildlife conservation on private lands in the Western Ghats. Pages
210-240 in G. Shahabuddin and M. Rangarajan, editors. Making
conservation work. Permanent Black, Ranikhet, India.
Nair, P. 2008. Agroecosystem management in the 21st century: it
is time for a paradigm shift. Journal of Tropical Agriculture
46:1-12.
Naughton-Treves, L., M. B. Holland, and K. Brandon. 2005. The
role of protected areas in conserving biodiversity and sustaining
local livelihoods. Annual Review of Environment and Resources
30(1):219-252. https://doi.org/10.1146/annurev.energy.30.050504.164507
Nayak, R., K. K. Karanth, T. Dutta, R. Defries, K. U. Karanth,
and S. Vaidyanathan. 2020. Bits and pieces: forest fragmentation
by linear intrusions in India. Land Use Policy: 104619. https://
doi.org/10.1016/j.landusepol.2020.104619
Nordén, A., J. Coria, A. M. Jönsson, F. Lagergren, and V. Lehsten.
2017. Divergence in stakeholders' preferences: evidence from a
choice experiment on forest landscapes preferences in Sweden.
Ecological Economics 132:179-195. https://doi.org/10.1016/j.
ecolecon.2016.09.032
Ojeda, D. 2012. Green pretexts: Ecotourism, neoliberal
conservation and land grabbing in Tayrona National Natural
Park, Colombia. Journal of Peasant Studies 39(2):357-375.
Peerlings, J., and N. Polman. 2009. Farm choice between agri-
environmental contracts in the European Union. Journal of
Environmental Planning and Management 52(5):593-612. https://
doi.org/10.1080/09640560902958131
Phalan, B., M. Onial, A. Balmford, and R. E. Green. 2011.
Reconciling food production and biodiversity conservation: Land
sharing and land sparing compared. Science 333(6047):1289-1291.
https://doi.org/10.1126/science.1208742
Pienaar, E. F., L. S. Jarvis, and D. M. Larson. 2014. Using a choice
experiment framework to value conservation-contingent
development programs: an application to Botswana. Ecological
Economics 98:39-48. https://doi.org/10.1016/j.ecolecon.2013.12.015
Pienaar, E. F., J. R. Soto, J. H. Lai, and D. C. Adams. 2019. Would
county residents vote for an increase in their taxes to conserve
native habitat and ecosystem services? Funding conservation in
Palm Beach County, Florida. Ecological Economics 159:24-34.
https://doi.org/10.1016/j.ecolecon.2019.01.011
Puri, M., A. Srivathsa, K. K. Karanth, I. Patel, and N. S. Kumar.
2020. The balancing act: maintaining leopard–wild prey
equilibrium could offer economic benefits to people in a shared
forest landscape of central India. Ecological Indicators 110:
105931. https://doi.org/10.1016/j.ecolind.2019.105931
Ramalingam, R., and P. Dharma Rajan. 2015. Needs for policy
on landscape restoration in India. Current Science 108
(7):208-1209.
Rathore, C. S., Y. Dubey, A. Shrivastava, P. Pathak, and V. Patil.
2012. Opportunities of habitat connectivity for Tiger (Panthera
tigris) between Kanha and Pench National Parks in Madhya
Pradesh, India. PLoS One 7(7): e39996. https://doi.org/10.1371/
journal.pone.0039996
Redpath, S. M., J. D. Linnell, M. Festa‐Bianchet, L. Boitani, N.
Bunnefeld, A. Dickman, R. J. Gutiérrez, R. J. Irvine, M.
Johansson, A. Majić, and B. J. McMahon. 2017. Don't forget to
look down—collaborative approaches to predator conservation.
Biological Reviews 92(4):2157-2163 https://doi.org/10.1111/
brv.12326
Robbins, P., A. Chhatre, and K. Karanth. 2015. Political ecology
of commodity agroforests and tropical biodiversity. Conservation
Letters 8(2):77-85. https://doi.org/10.1111/conl.12169
Roy, P. S., M. D. Behera, M. S. R. Murthy, A. Roy, S. Singh, S. P.
S. Kushwaha, C. S. Jha, S. Sudhakar, P. K. Joshi, C. S. Reddy et
al. 2015. New vegetation type map of India prepared using
satellite remote sensing: comparison with global vegetation maps
Ecology and Society 26(2): 46
https://www.ecologyandsociety.org/vol26/iss2/art46/
and utilities. International Journal of Applied Earth Observation
and Geoinformation 39:142-159. https://doi.org/10.1016/j.
jag.2015.03.003
Ruto, E., and G. Garrod. 2009. Investigating farmers’ preferences
for the design of agri-environment schemes: a choice experiment
approach. Journal of Environmental Planning and Management
52(5):631-647. https://doi.org/10.1080/09640560902958172
Schenk, A., M. Hunziker, F. and Kienast. 2007. Factors
influencing the acceptance of nature conservation measures—a
qualitative study in Switzerland. Journal of Environmental
Management 83(1):66-79. https://doi.org/10.1016/j.jenvman.2006.01.010
Scriven, J. 2012. Developing REDD+ policies and measures from
the bottom-up for the buffer zones of Amazonian protected areas.
Environment, Development and Sustainability 14(5):745-765.
https://doi.org/10.1007/s10668-012-9350-z
Sharma, A. 2017. Payment for ecosystem services. Economic and
Political Weekly 52(36):73-74.
Sharma, L. N., and O. R. Vetaas. 2015. Does agroforestry
conserve trees? A comparison of tree species diversity between
farmland and forest in mid-hills of central Himalaya. Biodiversity
and Conservation 24(8):2047-2061. https://doi.org/10.1007/
s10531-015-0927-3
Siebert, R., M. Toogood, and A. Knierim. 2006. Factors affecting
European farmers’ participation in biodiversity policies.
Sociologia Ruralis 46(4):318-340. https://doi.org/10.1111/
j.1467-9523.2006.00420.x
Sinha, B. C., Q. Qureshi, V. K. Uniyal, and S. Sen. 2012.
Economics of wildlife tourism—contribution to livelihoods of
communities around Kanha tiger reserve, India. Journal of
Ecotourism 11(3):207-218. https://doi.org/10.1080/14724049.2012.721785
Srivathsa, A., M. Puri, K. K. Karanth, I. Patel, and N. Samba
Kumar. 2019. Examining human–carnivore interactions using a
socio-ecological framework: sympatric wild canids in India as a
case study. Royal Society Open Science 6(5):182008. https://doi.
org/10.1098/rsos.182008
van den Berg, H. A., S. J. Riley, and S. L. Dann. 2011.
Conservation education for advancing natural resources
knowledge and building capacity for volunteerism. Society and
Natural Resources 24(3):205-220. https://doi.org/10.1080/08941
920902960404
West, P., J. Igoe, and Brockington, D. 2006. Parks and peoples:
the social impact of protected areas. Annual Review of
Anthropology 35:251-277. https://doi.org/10.1146/annurev.
anthro.35.081705.123308
Wikramanayake, E., E. Dinerstein, J. Seidensticker, S. Lumpkin,
B. Pandav, M. Shrestha, H. Mishra, J. Ballou, A. J. t. Johnsingh,
I. Chestin et al. 2011. A landscape-based conservation strategy
to double the wild tiger population. Conservation Letters 4
(3):219-227. https://doi.org/10.1111/j.1755-263X.2010.00162.x
Yumnam, B., Y. V. Jhala, Q. Qureshi, J. E. Maldonado, R. Gopal,
S. Saini, Y. Srinivas, and R. C. Fleischer. 2014. Prioritizing tiger
conservation through landscape genetics and habitat linkages.
PLoS One 9(11): e111207. https://doi.org/10.1371/journal.
pone.0111207
Appendix 1
D-efficiency - In the design of experiments, optimal designs are a class of experimental
designs that are optimal with respect to some statistical criterion (often related to the
variance-covariance matrix). Optimal designs allow parameters to be estimated without bias
and with minimum variance. A non-optimal design, in contrast, requires a greater number of
experimental runs to estimate the parameters with the same precision as an optimal design. A
D-optimal design is a computer generated design and consists of the best subset of
experiments selected from the full candidate set. For a given model, Y = Xβ + ε, with a D-
optimal design, the selected runs maximize the determinant of the information matrix X’X,
resulting in higher precision in the parameter estimates (Atkinson and Donev 1992).
Atkinson, A. C., and A. N. Donev. 1992. Optimum experimental designs. Oxford University
Press, Oxford.
Derivation of the reservation payment:
Respondent i has baseline indirect utility Ui0 from their status quo activities (i.e. current
farming practices), which is as a function of current household wealth (for example,
livestock, cash, and other storables) and current period income:
where the baseline utility for respondent i is Ui0, Ei is the households’s status quo wealth and
money income, εi0 is a random error, and γ is a parameter to be estimated. If respondent i
elects to enroll in agroforestry program j then their indirect utility function is:
where the utility derived from enrolling in agroforestry program j is Uij, Xij is a vector of the
non-monetary attributes of the program (e.g. land to be enrolled, duration of the program),
Paymentij is the annual payment per acre for enrollment in program j, εij is a random error,
and β is a vector of parameters to be estimated. Respondent i’s characteristics enter the model
as shifters on the β parameters.
The probability that respondent i chooses to enroll in agroforestry program j (i.e. they
respond that Yes they would prefer to enroll in that program) is:
where is the difference in errors and
(b)
(a)
is the utility difference between enrolling in agroforestry program j and not enrolling in the
program (i.e., remaining at status quo). The reservation price at which respondent i will
choose to enroll in agroforestry program j is the minimum payment (Paymentij) that
respondent i will accept for enrollment in the program, and is defined implicitly by the utility
difference being zero:
The reservation price can thus be solved for as:
Figure A1.1: Photographs used to illustrate the (a) current and (b) the envisioned state of the
land. These photographs were shown along with a detailed explanation of the kind of
agroforestry practices that the landowner could potentially undertake. Photographs are for
representation purpose only.
Photo credits:
Image (a) - https://www.flickr.com/photos/mckaysavage/2230560278
Image (b) – C Watson (https://blog.worldagroforestry.org/index.php/2016/05/23/in-
nicaragua-a-staggering-diversity-and-density-of-trees-on-farms/)
Table A1.1: Description of demographic, economic and socio-psychological variables and the parameters they interacted with in the two RPL
models.
Variable type
Abbreviation
Description
Parameter of
Interaction
Demographic
PDIS
Distance from Protected Area boundary
LAND, YEARS
Demographic
FCOV
Percentage forest cover in 1-km of landholding
LAND, YEARS
Demographic
CONF
Dummy variable takes a value of 1 if respondent has experienced conflict with
herbivore
LAND, YEARS
Demographic
AGE
Dummy variable takes a value of 1 if respondent more than 44 years (average
age)
LAND, YEARS
Demographic
EDU
Categorical variable for respondent's education
LAND, YEARS
Demographic
HHS
Household size
LAND, YEARS
Demographic
HIST
History/number of years living in the village
LAND, YEARS
Demographic
CAST
Caste of the respondent
LAND, YEARS
Economic
SIZE
Size of respondent's landholding
LAND, YEARS
Economic
AGIN
Income from agriculture
LAND, YEARS
Economic
OTIN
Income from other (non-agriculture) sources
LAND, YEARS
Economic
CROP
Number of crops grown
LAND, YEARS
Economic
SOUR
Number of income sources
LAND, YEARS
Socio-psychological
LINP
Component generated using PCA combining Likert statements relating to
additional requirements of effort and financial expenditure
ASC
Socio-psychological
LBNF
Component generated using PCA combining Likert statements relating to
provision of fodder, firewood, and additional income
ASC
Socio-psychological
LHWC
Likert statements relating to increase in conflict with animals
ASC
Socio-psychological
LSIZ
Likert response related to self-efficacy/worry about land size
ASC
Socio-psychological
LKNW
Likert response related to self-efficacy/worry about technical knowledge/training
ASC
Socio-psychological
LLBR
Likert response related to self-efficacy/worry about labor availability
ASC
Socio-psychological
LIRR
Likert response related to self-efficacy/worry about access to irrigation
ASC
Socio-psychological
LNBR
Likert response related to self-efficacy/worry about problems with neighbors
ASC
Socio-psychological
LSUC
Likert response related to self-efficacy/worry about trees not growing or fruiting
ASC
Socio-psychological
LFAM
Likert response related to social norms/ agreement with family members
ASC
Table A1.2: Demographic and socio-economic characteristics of respondents (N = 602)
Characteristics
Sub-characteristics
Details
Villages sampled
90
Number of people in HH1
(Average/HH)
5
Respondent
Household
Gender
Male
98%
43%
Female
2%
57%
Age
< 18
NA
30%
18-40
44%
46%
41-65
51%
19%
> 65
5%
5%
Education
illiterate
13%
19%
< 10th grade
53%
50%
10th grade
12%
10%
12th grade
10%
6%
Graduate and above
12%
14%
Caste
Scheduled Tribe
56%
Scheduled Caste
6%
Other Backward Class
31%
General
7%
History of living in village
0-5 years
0
6-19 years
1%
20-49 years
11%
> 50 years
88%
Livestock ownership -
Average/HH (min-max)
Cows
4.79 (0-34)
Buffaloes
1.27 (0-32)
Goat
1.55 (0-27)
Total
7.61 (0-66)
Landholding size
< 3 acres
4%
3-5 acres
38%
6-10 acres
30%
> 10 acres
28%
Number of crops grown -
Average/HH (min-max)
3 (1-5)
Annual Agricultural Revenue2 -
Median/HH (min-max)
Rs. 1,50,000 (~USD 2142)
(Rs. 10,000 – Rs. 30,00,000)
Non-agricultural Income Sources
Daily wage labor
44%
Dairy
9%
NTFP
56%
Non-service jobs3
3%
Service jobs4
26%
Business5
21%
Pension
11%
Number of non-agricultural
income sources - Average/HH
(min-max)
2 (0-5)
Annual Non-agricultural Income -
Median/HH
Rs. 51,200 (~USD 730)
Carnivore interaction (7 species6)
LP+HI+HD (personal loss)
12%
LP+HI+HD (in village)
70%
Herbivore interaction (crop
raiding) - number of species
named in top 3 problem animals
3 species
32%
2 species
32%
1 species
29%
None
7%
1 HH – household
2 Rs. – Indian Rupees (1 USD = Rs. 70)
3 Non-service jobs including employment in factory or as truck driver
4 Service jobs including employment as teacher, in tourism sector or government
5 Business including own shop, contractor, mill
6 Carnivores including tiger, leopard, bear, wolf, wild dog, jackal, fox; LP = Livestock
predation, HI – human injury, HD – human death
Table A1.3: Descriptive statistics for local context regarding rainfall trends, emigration,
forest dependence, and perceptions about living adjacent to forest, based on questionnaire
surveys (n = 602) in the buffer areas of Pench Tiger reserve, Madhya Pradesh, India.
Questions regarding local context
Categories
Percentage
Have you experienced changes in
rainfall pattern over the last 3 years
No change
3.32
Decreased
74.92
Increased
0.50
Erratic
18.44
Has the change in rainfall patterns
resulted in increase or decrease in
your agricultural income
No change
4.10
Decreased
95.56
Increased
0.34
What would you like for the future
of your family/children
Stay here and continue farming
22.84
Stay here but not farm
4.40
Move to city or town for job
60.41
They can decide for themselves
12.69
What resources do you obtain from
the forest
Firewood
62.46
Fodder
2.99
NTFP
49.17
Livestock grazing
49.38
Other
0.33
Do you like living near a forest
No
5.81
Yes
69.93
Neutral
24.25
Reasons for liking living near
forest
Ancestral ties, affiliation to land and
community support
82.01
Resource availability
36.86
Ecological and aesthetic value of forests
84.30
Job in tourism sector
5.64
Job in forest dept
0.88
Other reasons
7.94
Reasons for disliking living near
forest
Crop raiding by herbivores
87.36
Livestock loss to carnivores
23.08
Low standards of education
30.22
Poor healthcare
35.71
Limited job opportunities
20.33
Overcrowding due to tourism
1.10
Other reasons
21.43
Table A1.4: Estimated beta coefficients from multinomial logit (MNL) model incorporating
landowner characteristics of respondents from choice experiment surveys conducted in the
buffer area of Pench Tiger Reserve, India in 2018-19. Estimates are from the best-fit model
based on AIC.
Attribute
MNL
Coefficient
SE
ASC†
1.18***
0.07
Land
-0.04***
0.002
Year
-0.002
0.01
Payment
0.02***
0.002
BFCOV
0.11*
0.06
BCONF
0.14**
0.06
Model Properties
AIC/N
1.126
Log-Likelihood
-3035.44
BFCOV – binary variable with 1 = more than 25% forest cover in the 1-km buffer of
landholding
BCONF – binary variable with 1 = conflict (crop loss) with two or more herbivores
†ASC is an alternative specific constant taking the value 1 if one of the agroforestry programs
(Program A or Program B) is chosen and zero otherwise (Program C)
*** p < 0.01; ** p < 0.05; * 0.05 < p < 0.1
