Social Embeddedness as the Engine of Sustainable Food System: Evolutionary

Stable Strategies in Peri-Urban Agroforestry Systems

Abstract Peri-urban agroforestry systems constitute a strategic intersection for advancing food sovereignty, climate resilience,
and equitable development. However, the cooperative dynamics within these systems remain underexplored through evolutionary
game-theoretic approaches. This study investigates strategic interactions among low-to-middle-income farming households
engaged in coffee, fruit, and honey production in peri-urban settings where heightened climate vulnerability exacerbates income
inequality and food system fragility. Applying evolutionary game theory (EGT), we model how socially embedded cooperative
networks foster evolutionary stable strategies (ESS) that consistently outperform competitive equilibria across three empirically
measurable outcomes: (1) enhanced household food security through diversified organic production, (2) improved ecological
sustainability via biodiversity-mediated pest regulation and carbon sequestration, and (3) strengthened economic resilience via
circular waste valorization and premium market access. Analysis of over 100 cooperative farms in Southeast Asia reveals that
ESS adoption is catalyzed when social trust and shared infrastructure reduce transaction costs, enabling sustainable transitions
to organic certification. Our findings demonstrate that cooperative agroforestry emerges as a critical pathway for synergizing
climate adaptation, health outcomes, and poverty reduction within the UN Sustainable Development Goals framework. Extending
this analysis, we integrate empirically calibrated enhancements to agroforestry models across two additional major coffee-
exporting regions: Central/South America (Brazil) and Africa (Ethiopia). These regionally adapted models identify significant
opportunities for integrated agroforestry systems, particularly through coordinated alignment of Sub-national governance, bee
Apiaries, Fruit orchards, and coffEE farms (SAFE). Critically, current implementations in these regions fail to fully leverage
these synergies, revealing substantial potential for systemic optimization and sustainable intensification in global coffee
landscapes.

KEYWORDS: Sustainable food systems; Community supported agriculture; Organic; Sustainable food waste
management; Food security; Social embeddedness.

1. Introduction
Peri-urban agroforestry constitutes a strategically significant domain for sustainable development, where urban
expansion interfaces with agricultural production in ecologically complex landscapes requiring coordinated multi-
stakeholder governance. We analyze the adaptive decision-making dynamics among four key actors including sub-
nation governments, coffee farms, beekeeping operations, and fruit orchards, whose strategic interactions evolve
through feedback from relative payoffs, environmental stressors, and policy incentives. Despite the urgency of
developing resilient urban food systems and the established utility of EGT in modeling multi-actor systems, formal
analysis of this specific four-party interaction remains absent in the literature. This gap is particularly consequential
given the rapid global expansion of peri-urban agroforestry, now engaging over 800 million people in urban and
peri-urban agriculture [17]. The system’s inherent complexity, characterized by mutualistic ecological relationships,
resource competition, regulatory heterogeneity, and climate constraints, renders it uniquely suited for evolutionary
game-theoretic analysis to identify stable cooperative pathways.
Human Health
Organic food systems may contribute significantly to mitigating health challenges through nutrient-dense dietary
patterns. Current evidence indicates that diets rich in organic produce provide essential micronutrients such as
vitamins, minerals, antioxidants, and phytochemicals associated with enhanced tissue repair, metabolic regulation,
and immune function [79]. Observational studies suggest that such dietary patterns represent a key modifiable factor
in chronic disease prevention, potentially addressing underlying mechanisms like inflammation and oxidative stress
through whole-food, minimally processed sources [44]. Diets emphasizing phytonutrient-rich fruits, vegetables, and
whole grains show correlations with improved metabolic homeostasis, cognitive function, and longevity outcomes in
population-level analyses [28]. Regarding mental health, depression, frequently linked to chronic stress [49] may be
influenced by dietary components. Preliminary evidence indicates that honey consumption could modulate stress
responses, with rodent studies demonstrating reduced corticosterone levels under stress-induced conditions [52].
Human observational data further suggest associations between honey intake and lower markers of oxidative stress
and inflammation, though clinical evidence for direct effects on depression remains limited [33]. For cardiovascular
health, moderate coffee consumption, typically 3 to 4 cups a day, is associated with reduced risks of hypertension
and cardiovascular mortality in longitudinal cohort studies, challenging historical concerns about adverse
cardiovascular effects [10]; [50].

Economic Health
Climate change increasingly disrupts supply chains for coffee and honey through thermally driven ecological
stressors. Rising temperatures, precipitation variability, and pest proliferation threaten coffee cultivation, while
climate-induced pressures on pollinator populations compromise honey production [47]; [39]. These disruptions risk
diminishing access to nutrient-dense agricultural products with documented health associations, highlighting the
dependency of human well-being on ecological stability.

Brazil, accounting for 38% of global coffee exports, experienced severe climate-related production losses during the
2020-2021 growing season, when drought and frost affected 1.5 million hectares across Minas Gerais and São Paulo
states [14]. This event resulted in 600 million kg of unharvested coffee beans and $1.2 billion in direct economic
losses, disproportionately impacting 250,000 smallholder farms reliant on coffee income [14]. Similarly, Ethiopia’s
coffee sector which generates $1.1 billion annually from 400,000 tons produced by 5 million smallholders, faces
significant climate vulnerability [66]; [5]. Under high-emission scenarios, specialty coffee regions like Harar and
Yirgacheffe may lose more than 40% of climatically suitable land [11], jeopardizing premium market access and
threatening livelihoods for farmers dependent on price premiums for specialty varieties.

Vietnam’s coffee sector further illustrates the nexus of climate exposure and agricultural sustainability. In the Central
Highlands were producing 95% of Vietnam’s coffee, however, drought and erratic rainfall have reduced yields by
10% to 15% since 2015, while excessive precipitation intensifies coffee berry disease and root rot incidence [25].
These climate impacts compound pre-existing degradation from monoculture practices, agrochemical overuse, and
deforestation, which have diminished soil organic carbon in key regions [71]. Concurrently, salinity intrusion in the
Mekong Delta, affecting 75% of Vietnam’s rice production, exacerbates water scarcity through sea-level rise and
groundwater depletion [34], creating cross-commodity vulnerabilities. These systemic risks underscore the
imperative for integrated adaptation strategies, including agroforestry diversification and precision water
management, to enhance supply chain resilience while maintaining ecological integrity.

National Health
Small and medium-scale family farms dominate peri-urban coffee production, representing 80% to 90% of global
coffee-growing households [16]. Climate vulnerability exacerbates pre-existing socioeconomic inequalities in this
sector, as resource-constrained smallholders face disproportionate exposure to climate-induced yield losses.
Specifically, limited access to formal credit, climate-resilient infrastructure, and adaptive technologies impede
drought and flood mitigation capacity [83]. In contrast, larger agribusinesses and export cooperatives equipped with
capital for irrigation, crop diversification, and formal risk management tools demonstrate greater capacity to absorb
production shocks. Consequently, the growing marginalization of smallholders in climate-vulnerable markets
threatens export stability. These dynamics underscore the critical need to integrate equity dimensions into climate
adaptation frameworks for sustainable agricultural systems.

Agroforestry - solution for optimal productivity

Agroforestry emerges as the optimal solution for agriculture relied on countries because it simultaneously addresses
climate vulnerability, income inequality, and biodiversity loss through a synergistic, low-input system. By
integrating coffee plants with fruit trees such as durian, mango, or lychee and establishing managed apiaries for
honeybees, this model creates a resilient agricultural ecosystem and a sustainable food system. Crucially, bees
thrive in this diverse setting, fruit trees provide continuous nectar and pollen sources during the dry season when
coffee blossoms are scarce, ensuring year-round bee activity and pollination efficiency [81]. This directly boosts
coffee yield as bees increase fruit yield around 20% [76] while simultaneously diversifying farm income: the fruit
trees generate high-value cash crops, and the honey, produced from a richer, more varied floral palette, commands
premium prices in eco-conscious markets. For smallholder farmers, this intercropping system reduces reliance on
volatile coffee prices, stabilizes household income, and minimizes soil erosion, all while enhancing carbon
sequestration. Thus, the SAFE model transforms climate threats into economic opportunity. Allow families to have a
small farmer’s market with diverse products at home, turning marginalized farmers into dual-income producers
without requiring costly external inputs, proving that ecological harmony and equitable prosperity are not just
possible, but interdependent, strengthen the social embeddedness.

The coffee-fruit intercropping model fundamentally transforms agricultural ecology by replacing chemical
dependency with biodiversity-driven resilience, directly enabling a scalable transition to organic certification. By
integrating fruit trees with coffee, a system where complementary root systems and floral diversity create a self-
regulating ecosystem, pest pressure is drastically reduced: diverse plant species disrupt pest life cycles as coffee leaf
miners avoid fruit tree allelopathic compounds [42], while honeybees and beneficial insects naturally control aphids
and mites without synthetic pesticides. This eliminates the need for broad-spectrum insecticides, reducing chemical
runoff into waterways and rebuilding soil microbiome health through organic matter retention [15]. Crucially, the
model meets organic certification standards by design: the absence of pesticides, coupled with nutrient symbiosis,
satisfies organic soil fertility requirements while producing certifiable organic honey from diverse nectar sources. In
Vietnam, this can be verified by VietGap, an organic verified organization which can allow food labeling for family
farmers [98]. This indicates this model as the most efficient pathway to profitable organic transition for
smallholders, turning climate vulnerability into a catalyst for regenerative agriculture.

The coffee-fruit intercropping model synthesizes circular and sharing economies into a self-reinforcing system
where waste becomes shared wealth, eliminating both environmental harm and economic inequality. Farmers can
collect coffee pulp, fruit peels, and overripe produce into a circular nutrient loop: this waste is transformed via shared
solar-powered composting units (cooperatively funded) into certified organic fertilizer, which is then distributed
freely across the network to all member farms, creating a sustainable food waste management [80]. This dual
mechanism eliminates significant synthetic fertilizer use with empirical validation in Vietnam as several coffee farms
have not utilized fertilizer for more than 20 years, reducing pollution while generating additional shared revenue
proving that waste is not an end, but the beginning of a shared cycle [100]. This is not just circularity, its collective
resilience, where every discarded piece fuels the next harvest, the next income, and the next generation, enhancing
the community supported agriculture. Since the model utilized coffee as one of the main components, this model
is visible within communities including Central and South America, Africa and The Middle East, and Southeast Asia
[75].

Sustainable Development Goals

The SAFE agroforestry model delivers transformative synergies across multiple Sustainable Development Goals
(SDGs). By diversifying farm income through coffee, high-value fruits and premium honey, the model directly
advances SDG 1 (No Poverty) and SDG 10 (Reduced Inequalities), lifting smallholder families out of income
volatility while narrowing the gap between resource-rich agribusinesses and marginalized farmers [103].
Simultaneously, it strengthens SDG 2 (Zero Hunger) through enhanced soil fertility, reduced erosion, and year-round
food security from diverse crops [103]. Crucially, the system embodies SDG 13 (Climate Action) by sequestering
carbon in multi-layered canopies and using bees to maintain pollination resilience during droughts [103]. SDG 15
(Life on Land) is advanced via biodiversity conservation: diverse flowering trees sustain bee populations, boosting
coffee pollination efficiency by 20% while producing higher-quality honey for ethical markets which also meets
SDG 12 (Responsible Consumption) [76]; [103]. This holistic approach proves that ecological regeneration and
equitable prosperity are inseparable, turning climate vulnerability into a catalyst for SDG convergence, where every
bee, tree, and farmer becomes a pillar of sustainable development.

2. Literature Review
Critical Analysis of Current Limitations
2.1 Temporal Dynamics Deficiency in Agricultural Modeling
Contemporary agricultural policy frameworks predominantly rely on static equilibrium models that fundamentally
misrepresent the temporal complexity of agroecological systems. While static Computable General Equilibrium
models provide snapshot analyses of economic states, they inherently fail to capture evolutionary processes essential
for sustainable agriculture planning [104]. This temporal limitation proves particularly problematic in pollinator-
dependent crop systems where cause-effect relationships operate through recursive feedback loops rather than linear
causality. Crucially, integrated crop-livestock systems (ICLS) evaluation increasingly employs multi-criteria
assessment frameworks based on objective analysis rather than subjective metrics yet still lacks dynamic modeling of
temporal evolution across trophic levels [105]. The absence of dynamics in current approaches results in policy
recommendations that are fundamentally disconnected from the evolutionary reality of agricultural ecosystems.

2.2 The Heterogeneous Agent Gap in Circular Bioeconomy Modeling

Despite growing recognition of circular bioeconomy principles, existing economic frameworks fail to adequately
model strategic interactions among heterogeneous stakeholders with asymmetric power dynamics. Current literature
demonstrates limited capacity to represent how government regulatory interventions influence multi-agent strategy
evolution in agricultural contexts, particularly regarding non-cooperative equilibria between state actors and
ecological agents [106]. While EGT has been successfully applied to government, farmer and enterprise interactions
game models [107], a critical void remains in modeling quadruple-helix interactions that integrate state regulators,
pollinator populations, and multiple crop systems within a unified theoretical framework. This deficiency is
particularly acute in multi-trophic agricultural systems where frequency-dependent selection processes drive co-
evolutionary dynamics between species and human actors. Brazil and Ethiopia remain trapped in a critical
agricultural blind spot: while Brazil pioneers’ coffee-fruit intercropping such as coffee-jaboticaba-guava and
Ethiopia relies on coffee-shade tree or coffee-food crop systems, both fail to address the dry-season pollinator
collapse that devastates yields.

2.3 Systemic Research Bias in Agroforestry

Global agroforestry research exhibits a critical bias toward fruit-fruit intercropping systems, systematically
overlooking the transformative potential of industrial crop-fruit crop integration for smallholder resilience, despite its
urgent relevance in climate-vulnerable regions. This gap stems from three flawed assumptions: the erroneous belief
that industrial crops like coffee inherently disrupt biodiversity, ignoring their complementary nutrient cycles with
fruit trees, an overemphasis on monoculture efficiency over systemic income stability, and a failure to address the
economic reality that smallholders require both short-term cash (from coffee) and long-term revenue (from fruit
trees) to withstand price volatility. Consequently, studies neglect to analyze seasonal phenological mismatches, bee-
driven honey quality premiums and socioeconomic adoption barriers like land tenure insecurity.

2.4 Empirical Validation Deficits in Multi-Species Modeling

Theoretical advancements in multi-species interaction modeling remain largely disconnected from empirical
validation in complex agricultural settings. Although EGT provides robust mechanisms for analyzing strategic
options and behavioral change among stakeholders [108], current applications lack field-validated parameterization
for integrated coffee-fruit agroecosystems with pollinator dependencies. Recent methodological innovations in
system dynamics evolutionary game modeling show promise for multi-agent cooperation analysis [109] yet fail to
incorporate the specific ecological constraints of tropical highland farming systems where coffee and fruit trees share
pollination resources. This creates a significant translational gap between theoretical models and on-farm
implementation potential.

Novel Contribution of the Proposed Framework

This study addresses these interconnected limitations through the development of the SAFE model evolutionary
game framework, which makes three substantive contributions to literature:

We operationalize EGT's replicator dynamics to model strategy evolution across 15 distinct ecological-economic
scenarios, explicitly capturing how government policy interventions trigger cascading behavioral adaptations across
trophic levels. Unlike static approaches, our framework demonstrates how initial cooperative equilibria may
destabilize through frequency-dependent selection processes under changing environmental conditions

The model innovatively incorporates government as a meta-strategic actor whose regulatory instruments alter the
payoff structure for all other agents, thereby modeling power dynamics absent in conventional game-theoretic
approaches. This advances beyond existing government-farmer-enterprise models [110] by integrating ecological
agents (bees) as strategic players with evolutionary fitness considerations, creating a novel quadruple-helix analytical
structure. Besides that, the model tackles the main core of agriculture which is plants and animals rather than farmers
or enterprises.

The SAFE agroforestry model directly bridges the research gap by delivering quantifiable socioeconomic and
ecological co-benefits through three integrated innovations. First, it establishes income dualism: coffee provides
immediate cash flow with 3 to 6 months post-harvest [54], while fruit trees generate long-term revenue with 5 to 10
years required for harvest [20], shielding smallholders from market volatility pilot farms in Vietnam’s Central
Highlands. It also ensures both a short and long food supply chain. Second, it leverages nutrient symbiosis: coffee’s
nitrogen demands, and fruit trees’ potassium/magnesium needs create complementary root systems that reduce
fertilizer through ecological niche partitioning, reversing soil degradation [101]. Third, it unlocks high-value end-
product diversification, transforming raw outputs into market-ready goods like honey-infused coffee bars, fruit-
coffee herbal teas, zero-waste coffee husk snacks, and healthy cereal such as Granola [95]; [77]; [53]; [48].
Crucially, bees sustain year-round pollination during coffee’s dry season via diverse fruit crops nectar, boosting
coffee yields by 20% while producing premium honey [76]. In Vietnam, these end-products can be verified by a
third-party called VietGap to ensure the organic and allow food labeling for the product, which enhances the trust
from customers [98]. This model democratizes access to nutrient-dense food such as low-sugar cereal for elders or
fruit ice cream with nuts for children which enhances food security and proves ecological integration is the scalable
solution to turn climate vulnerability into equitable prosperity for smallholder farming families.
Leveraging primary data from over 100 certified coffee-fruit farms in Vietnam's Central Highlands, we calibrate
model parameters with empirical validation. This addresses the critical validation gap identified in current
agricultural game theory applications, where models often lack connection to real-world behavioral data.

Theoretical Positioning and Novelty

Our framework specifically bridges two critical divides in the literature: (1) between ecological modeling that
captures multi-species interactions and economic modeling that addresses strategic decision-making under bounded
rationality [107]; and (2) between theoretical game structures and empirically validated behavioral parameters. While
recent scholarship has begun exploring evolutionary game mechanisms for agricultural sustainability [111], no
existing work simultaneously incorporates government regulatory power, pollinator ecology, and multi-crop
economics within a single dynamic framework. The SAFE model thus represents a methodological advancement
beyond current models by formalizing the co-evolutionary dynamics between policy instruments, ecological
services, and agricultural production strategies through rigorous EGT.

This research directly responds to the urgent need identified in sustainability science for dynamic models that
demonstrate the integration in agriculture for long-term sustainability [112], while providing the empirical grounding
necessary for actionable policy insights in coffee-fruit agroecosystems globally.

2.4 Methodology
The SAFE framework distinguishes itself through its unprecedented synthesis of dynamic economic modeling,
disciplinary agricultural expertise, and field-validated parameterization, a triad rarely achieved in contemporary
sustainability science. Unlike conventional economic models that treat agricultural systems as exogenous production
functions, our approach embeds granular agroecological knowledge directly into the game-theoretic architecture:
pollination efficiency curves for Apis cerana in Coffea arabica-fruit tree mosaics, phenological synchronization
constraints between coffee flowering and fruit tree blooming seasons, and soil nutrient cycling dynamics in volcanic
highland agroecosystems [62]. Critically, this theoretical structure is not merely illustrative but empirically grounded
through primary data collection from around 100 certified farms across Vietnam's Central Highlands where they
produce 95% of the nation's coffee [32]. By statistically calibrating payoff matrices using measured variables, we
transform abstract evolutionary game constructs into a contextually precise simulation engine

3. Hypothesis
Hypothesis 1: Players
This evolutionary game model involves four primary stakeholders: the Sub-national governance, Fruit orchards, Bee
apiaries and Coffee farms. All participants are assumed to be rational, risk-neutral decision-makers operating under
conditions of complete and perfect information. The Subnational environmental governance faces a binary strategic
choice between enforcing “Environmental Regulation” or allowing “Lax Regulation.” Meanwhile, the three
corporate actors each choose between engaging in “Resource Sharing and Waste Management” or opting out with
“No Resource Sharing and No Waste Management.” The strategic interplay among these agents drives the system’s
evolutionary dynamics, shaping the emergence, stability and long-term viability of cooperative behavior.

Hypothesis 2: Strategies
Let 𝑣 denote the probability that the Subnational Environmental Governance adopts the “Environmental Regulation”
strategy, while 1 − 𝑣 represents the probability of selecting the “Lax Regulation” strategy. Similarly, the probability
that Fruit Orchards adopt the “Resource Sharing and Waste Management” strategy is 𝑥, with 1 − 𝑥 representing the
alternative. Bee Apiaries choose the cooperative strategy with probability 𝑦 and the non-cooperative strategy with
probability 1 − 𝑦. The probability the Coffee farms choose ‘Resource Sharing and Waste Management’ strategy is 𝑧,
while the probability of ‘No Resource Sharing and No Waste Management’ strategy is 1 − 𝑧. These probability
parameters reflect the dynamic strategic preferences of each participant within the evolutionary framework.

Hypothesis 3: Reputational
When the Subnational Environmental Governance enforces environmental regulations, it reinforces its public image
by signaling effective governance, a strong commitment to sustainability and active oversight of agricultural
cooperatives. This regulatory stance enhances its institutional credibility in promoting resource efficiency and
environmental responsibility. Conversely, adopting a lax regulatory approach exposes the institution to reputational
risks, as it may be perceived as ineffective in managing sustainable agricultural practices.

Reputational dynamics similarly affect the Fruit Orchards, Bee Apiaries and Coffee Farms. By engaging in the
“Resource Sharing and Waste Management” strategy, these enterprises not only strengthen their public image but
also gain competitive advantages, expand market share and position themselves to benefit from favorable policy
incentives. However, if the governance body maintains a lax stance and any of the three corporations opts out of
cooperative practices, the regulator faces heightened reputational costs. Its failure to enforce compliance can
undermine public trust, weaken institutional legitimacy and reduce the overall effectiveness of sustainability policies.

Hypothesis 4: Incentives and Sanctions

When the Subnational Governance enforces environmental regulations, it introduces a suite of incentives to promote
adoption of the “Resource Sharing and Waste Management” strategy among Fruit Orchards, Bee Apiaries and
Coffee Farms. These incentives include direct financial support for sustainable practices such as intercropping,
organic fertilizer production and waste recycling, thereby reducing operational costs. In addition, technical assistance
is offered through training programs, access to green technologies and knowledge-sharing platforms aimed at
improving environmental performance. Market-based incentives such as preferential access to government
procurement, green certifications and tax benefits further enhance the competitiveness of cooperative enterprises.
Collectively, these measures are designed to catalyze a self-sustaining circular economy in which environmentally
responsible behaviors are both economically viable and scalable.
Conversely, non-cooperative actors that reject the “Resource Sharing and Waste Management” strategy face a
structured set of sanctions. These include financial penalties for inefficient resource use and improper waste disposal,
as well as increased compliance costs through intensified inspections, expanded reporting requirements and
administrative burdens. Access to public support mechanisms such as subsidies, trade incentives and certification
programs may also be restricted. Furthermore, the public disclosure of non-compliance can erode consumer trust and
investor confidence, damaging reputations. Through these enforcement mechanisms, the Subnational Environmental
Governance seeks to deter unsustainable practices and ensure the resilience and stability of cooperative
environmental strategies over the long term.

Figure 1: Impact of Government to Agriculture corporates

Hypothesis 5: Additional profits

Adoption of the “Resource Sharing and Waste Management” strategy yields additional economic benefits for Fruit
Orchards, Bee Apiaries and Coffee Farms, as illustrated in Figure 2.
 Figure 2: Benefit contribution between corporates

Fruit Orchards derive additional profits by intercropping with coffee, selling organic fertilizers produced from fruit
waste to Coffee Farms, and purchasing low-cost organic fertilizers derived from coffee by-products. Beyond
economic gains, Fruit Orchards also benefit environmentally through improved soil health and reduced chemical
input. Collaboration with Bee Apiaries further enhances pollination efficiency, particularly for coffee trees within
intercropping systems, thereby increasing overall productivity.

Bee Apiaries benefit from an abundant and diverse nectar supply provided by both coffee and fruit trees, supporting
healthier colonies and potentially higher honey yields.

Similarly, Coffee Farms gain economic advantages from intercropping fruit trees, marketing organic fertilizers made
from coffee waste to Fruit Orchards, and acquiring low-cost organic fertilizers produced from fruit residues. These
practices also contribute to environmental sustainability. As with Fruit Orchards, cooperation with Bee Apiaries
improves the pollination of fruit trees, enhancing productivity within the intercropped system.
 Figure 3: Pollination plan within a year between fruits and coffee tree

Intercrop model explanation

As previously discussed, Bee Apiaries serve as a critical support actor for both fruit and coffee production through
their role in pollination. Several bees can be found in the Central Highland with significant contributions in
pollination, such as Apis mellifera and Apis cerana [119]. However, to fully optimize pollination efficiency and
ensure a stable food supply for bees year-round, the model requires a carefully designed fruit tree cultivation plan. A
key ecological condition is the simultaneous blooming of different plants to support cross-pollination and maintain
continuous nectar availability. There can be numerous combinations between coffee and fruits, however for detailed
analysis, the paper gives a model based on actual data in Dak Lak developed from the existing models [74]; [85];
[58]; [13]; [36]; [31]; [24]; [89]; [27]; [37]; [88]; [30].

Figure 3 presents an example of such a planting schedule tailored for the Tay Nguyen region.
From January to April, Coffee and Durian (Ri6; Dona) [120] will be the main contributors while avocado (Fuerte)
[121] plays a role as a subordinate in case coffee (Robusta) [113] flowering is delayed due to dry season [62]; [26];
[90]; [18]. For April to July, Durian continues to be the main contributor of pollen and nectar with Pepper (Piper
nigrum) [122] while Banana (Musa cavendishii) [123] is the sub-source [90]; [61]; [19]. From July to October,
Pepper and Banana are the main producers. From October to December, Avocado and Jackfruit play the main role in
being a supplier for bees while durian is the side supplier [65]; [35]. However, durian during this time is off-season,
not the main flower period, however, it is still possible for farmers to cultivate to provide flowers with several
techniques such as watering or flower stimulation fertilizer [99]. This model is verified by previous successful
related models, showing no nutrient rival while helping each other in terms of pesticide elimination and nutrient
sharing.

A strategic rationale behind the selection of specific fruit tree species lies in their significant roles in Vietnam’s
agricultural export sector and their strong global market rankings. Vietnam has emerged as a major exporter of
several high-value fruits. In 2023, Vietnam became the leading durian supplier to China, surpassing Thailand, with
export volumes reaching 500,000 tons valued at USD 2.3 billion [115]. Furthermore, Vietnam was ranked the third
largest pepper exporter in 2020 which is valued at nearly 800 million USD [69]. Besides that, Vietnam is the fourth
largest coffee supplier [68]. Vietnam banana was ranked 12th in the world exporter in 2023 [67]. For avocado, the
potential for this market is higher than ever with the support from the Government to export to the largest
consumption market such as the USA, China or Australia [116]; [117]; [118]. Thus, this indicates that this model is
not only optimal in terms of integration, it also has a strong connection with Vietnam export advantage.

In summary, the complete set of parameter definitions and their corresponding interpretations are presented in Tables
1 to Table 4.

Sub-national governance

v The probability of Subnational environmental

governance choosing the 'Environmental Regulation”
strategy.

(1-v) The probability of Subnational environmental

governance choosing the 'Lax Regulation” strategy.

D Reputational benefits for Subnational environmental

governance enforcing environmental regulations.

E Reputational risks for Subnational environmental

governance when regulations are lax and parties choose
the 'No Resource Sharing and No Waste Management”
strategy.

F Damage to parties when not receiving support from

Subnational environmental governance.

G Incentives from Subnational environmental governance

when parties choose the 'Resource Sharing” strategy.

H Incentives from Subnational environmental governance

when parties choose the 'Waste Management” strategy.

I Sanctions by Subnational environmental governance

when parties choose the 'No Resource Sharing” strategy.

J Sanctions by Subnational environmental governance

when parties choose the 'No Waste Management”
strategy.

Table 1: Parameters of Subnational Environmental Governance

 Fruit orchards (a)

x The probability of Fruit orchards choosing the

'Resource Sharing and Waste Management” strategy.

(1-v) The probability of Fruit orchards choosing the no

'Resource Sharing and Waste Management” strategy.

𝐷𝑎 Reputational benefits for Fruit orchards enforcing

environmental regulations.

𝐾𝑐𝑎 Additional profit for Fruit orchards intercropping with

the Coffee farms.

𝐿𝑎𝑐 Additional profit for Fruit orchards by trading organic

fertilizer with the Coffee farms.

𝑂𝑐𝑎 Additional profit for Fruit orchards as environmental

benefit with the Coffee farms.

𝑀𝑏𝑎 Additional profit for Fruit orchards by pollination of

the Bee apiaries.

Table 2: Parameters of Fruit orchards

Bee apiaries (b)

y The probability of Bee apiaries choosing the 'Resource

Sharing and Waste Management” strategy.

(1-y) The probability of Bee apiaries choosing the no

'Resource Sharing and Waste Management” strategy.

𝐷𝑏 Reputational benefits for Bee apiaries enforcing

environmental regulations.

𝐾𝑎𝑏 Additional profit for Bee apiaries as an abundant source

 of nectar from Fruit orchards.

𝐿𝑏𝑐 Additional profit for Bee apiaries as an abundant source

of nectar from Coffee farms.

Table 3: Parameters of Bee apiaries

Coffee farms (c)

z The probability of Coffee farms choosing the 'Resource

Sharing and Waste Management” strategy.

(1-z) The probability of Coffee farms choosing the no

'Resource Sharing and Waste Management” strategy.

𝐷𝑐 Reputational benefits for Coffee farms enforcing

environmental regulations.

𝐾𝑎𝑐 Additional profit for Coffee farms intercropping with

the Fruit orchards.

𝐿𝑐𝑎 Additional profit for Coffee farms by trading organic

fertilize with the Fruit orchards.

𝑂𝑎𝑐 Additional profit for Coffee farms as environmental

benefit with the Fruit orchards.

𝑀𝑏𝑐 Additional profit for Coffee farms by pollination of the

Bee apiaries.

Table 4: Parameters of Coffee farms

4. Payoff matrix and Tree diagram

Based on the established hypotheses and defined parameters, a four-party evolutionary game payoff matrix involving
SAFE is constructed. The complete payoff matrix is presented in Table 5.

Subnational environmental governance (Environmental Regulation) (𝑣)

 a Resource Sharing and Waste No Resource Sharing and No Waste
Management (𝑥) Management (1 − 𝑥)
b

Resource 𝐷 − 𝐺 𝑎 − 𝐻𝑎 − 𝐺 𝑏 − 𝐻𝑏 − 𝐺 𝑐 − 𝐻𝑐 𝐷 + 𝐼𝑎 + 𝐽𝑎 − 𝐺 𝑏 − 𝐻 𝑏 − 𝐺 𝑐 − 𝐻 𝑐 Resource
𝑎 𝑎
𝐷 +𝐺 +𝐻 + 𝑎
𝐾𝑐𝑎 + 𝐿𝑎𝑐 + 𝑂𝑐𝑎 + 𝑀𝑏𝑎 −𝐷𝑎 − 𝐼𝑎 − 𝐽𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎
Sharing and Sharing
𝐷𝑏 + 𝐺 𝑏 + 𝐻 𝑏 + 𝐾𝑎𝑏 + 𝐿𝑏𝑐 𝐷𝑏 + 𝐺 𝑏 + 𝐻 𝑏 − 𝐾𝑎𝑏 + 𝐿𝑏𝑐
Waste and
𝐷𝑐 + 𝐺 𝑐 + 𝐻 𝑐 + 𝐾𝑎𝑐 + 𝐿𝑐𝑎 + 𝑂𝑎𝑐 + 𝑀𝑏𝑐 𝐷𝑐 + 𝐺 𝑐 + 𝐻 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 + 𝑀𝑏𝑐
Management Waste
(𝑧) Management
(𝑦)
No Resource 𝐷 − 𝐺 𝑎 − 𝐻 𝑎 − 𝐺 𝑏 − 𝐻 𝑏 + 𝐼𝑐 + 𝐽𝑐 𝐷 + 𝐼𝑎 + 𝐽𝑎 − 𝐺 𝑏 − 𝐻 𝑏 + 𝐼𝑐 + 𝐽𝑐
𝐷𝑎 + 𝐺 𝑎 + 𝐻 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 + 𝑀𝑏𝑎 −𝐷𝑎 − 𝐼𝑎 − 𝐽𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎
Sharing and No
𝑏 𝑏 𝑏
𝐷 +𝐺 +𝐻 + 𝐾𝑎𝑏 − 𝐿𝑏𝑐 𝐷𝑏 + 𝐺 𝑏 + 𝐻 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐
Waste
−𝐷𝑐 − 𝐼𝑐 − 𝐽𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 −𝐷𝑐 − 𝐼𝑐 − 𝐽𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐
Management
(1 − 𝑧)

Resource 𝐷 − 𝐺 𝑎 − 𝐻 𝑎 + 𝐼𝑏 + 𝐽𝑏 − 𝐺 𝑐 − 𝐻 𝑐 𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐼𝑏 + 𝐽𝑏 − 𝐺 𝑐 − 𝐻 𝑐 No Resource
𝑎 𝑎
𝐷 +𝐺 +𝐻 + 𝑎
𝐾𝑐𝑎 + 𝐿𝑎𝑐 + 𝑂𝑐𝑎 − 𝑀𝑏𝑎 −𝐷𝑎 − 𝐼𝑎 − 𝐽𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎
Sharing and Sharing
−𝐷𝑏 − 𝐼𝑏 − 𝐽𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 −𝐷𝑏 − 𝐼𝑏 − 𝐽𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐
Waste and
𝐷𝑐 + 𝐺 𝑐 + 𝐻 𝑐 + 𝐾𝑎𝑐 + 𝐿𝑐𝑎 + 𝑂𝑎𝑐 − 𝑀𝑏𝑐 𝐷𝑐 + 𝐺 𝑐 + 𝐻 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐
Management No Waste
(𝑧) Management
(1 − 𝑦)
No Resource 𝐷 − 𝐺 𝑎 − 𝐻 𝑎 + 𝐼𝑏 + 𝐽𝑏 + 𝐼𝑐 + 𝐽𝑐 𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐼𝑏 + 𝐽𝑏 + 𝐼𝑐 + 𝐽𝑐
𝐷𝑎 + 𝐺 𝑎 + 𝐻 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 −𝐷𝑎 − 𝐼𝑎 − 𝐽𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎
Sharing and No
𝑏 𝑏 𝑏
−𝐷 − 𝐼 − 𝐽 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 −𝐷𝑏 − 𝐼𝑏 − 𝐽𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐
Waste
−𝐷𝑐 − 𝐼𝑐 − 𝐽𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 −𝐷𝑐 − 𝐼𝑐 − 𝐽𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐
Management
(1 − 𝑧)

Table 5: Payoff matrix of Fruit orchards, Bee apiaries and Coffee farms when Subnational environmental
governance choose Environmental Regulation strategy

Subnational environmental governance (Lax Regulation) (1 − 𝑣)

a Resource Sharing and Waste No Resource Sharing and No Waste

Management (𝑥) Management (1 − 𝑥)
b
c
 Resource −𝐷 −𝐷 − 𝐸 𝑎 Resource
𝑎
𝐷 −𝐹 + 𝑎
𝐾𝑐𝑎 + 𝐿𝑎𝑐 + 𝑂𝑐𝑎 + 𝑀𝑏𝑎 −𝐷 − 𝐹 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎
𝑎 𝑎
Sharing and Sharing
𝐷𝑏 − 𝐹 𝑏 + 𝐾𝑎𝑏 + 𝐿𝑏𝑐 𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 + 𝐿𝑏𝑐
Waste and
𝐷𝑐 − 𝐹 𝑐 + 𝐾𝑎𝑐 + 𝐿𝑐𝑎 + 𝑂𝑎𝑐 + 𝑀𝑏𝑐 𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 + 𝑀𝑏𝑐
Management Waste
(𝑧) Management
(𝑦)
No Resource −𝐷 − 𝐸 𝑐 −𝐷 − 𝐸 𝑎 − 𝐸 𝑐
𝐷𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 + 𝑀𝑏𝑎 −𝐷𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎
Sharing and No
𝑏 𝑏
𝐷 −𝐹 + 𝐾𝑎𝑏 − 𝐿𝑏𝑐 𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐
Waste
−𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 −𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐
Management
(1 − 𝑧)

Resource −𝐷 − 𝐸 𝑏 −𝐷 − 𝐸 𝑎 − 𝐸 𝑏 No Resource
𝑎
𝐷 −𝐹 + 𝑎
𝐾𝑐𝑎 + 𝐿𝑎𝑐 + 𝑂𝑐𝑎 − 𝑀𝑏𝑎 −𝐷 − 𝐹 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎
𝑎 𝑎
Sharing and Sharing
−𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 −𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐
Waste and
𝐷𝑐 − 𝐹 𝑐 + 𝐾𝑎𝑐 + 𝐿𝑐𝑎 + 𝑂𝑎𝑐 − 𝑀𝑏𝑐 𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐
Management No Waste
(𝑧) Management
(1 − 𝑦)
No Resource −𝐷 − 𝐸 𝑏 − 𝐸 𝑐 −𝐷 − 𝐸 𝑎 − 𝐸 𝑏 − 𝐸 𝑐
𝐷𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 −𝐷𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎
Sharing and No
𝑏 𝑏
−𝐷 − 𝐹 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 −𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐
Waste
−𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 −𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐
Management
(1 − 𝑧)

Table 6: Payoff matrix of Fruit orchards, Bee apiaries and Coffee farms when Subnational environmental
governance choose Lax Regulation strategy

The payoff matrix can also be illustrated in a tree diagram which is provided in Figure 4 and Figure 5.
 Figure 4: Tree diagram when Subnational Environmental Governance chooses Environmental Regulation strategy

Figure 5: Tree diagram when Subnational Environmental Governance chooses Lax Regulation strategy

5. Expected Utility and Replicator Dynamic

Based on the four-party evolutionary game payoff matrix, the paper applied Expected Utility strategy from game
theory to calculate each player's strategies payoff.

a, Subnational environmental governance

The expected utility of Subnational environmental governance choosing the 'Environmental Regulation” strategy is:
𝐸𝑈𝑣𝐺 = 𝑥𝑦𝑧(𝐷 − 𝐺 𝑎 − 𝐻𝑎 − 𝐺 𝑏 − 𝐻𝑏 − 𝐺 𝑐 − 𝐻𝑐 ) + (1 − 𝑥)𝑦𝑧(𝐷 + 𝐼 𝑎 + 𝐽𝑎 − 𝐺 𝑏 − 𝐻𝑏 − 𝐺 𝑐 − 𝐻𝑐 ) +
𝑥𝑦(1 − 𝑧)(𝐷 − 𝐺 𝑎 − 𝐻𝑎 − 𝐺 𝑏 − 𝐻𝑏 + 𝐼 𝑐 + 𝐽𝑐 ) + (1 − 𝑥)𝑦(1 − 𝑧)(𝐷 + 𝐼 𝑎 + 𝐽𝑎 − 𝐺 𝑏 − 𝐻𝑏 + 𝐼 𝑐 + 𝐽𝑐 ) +
𝑥(1 − 𝑦)𝑧(𝐷 − 𝐺 𝑎 − 𝐻 𝑎 + 𝐼 𝑏 + 𝐽𝑏 − 𝐺 𝑐 − 𝐻𝑐 ) + (1 − 𝑥)(1 − 𝑦)𝑧(𝐷 + 𝐼 𝑎 + 𝐽𝑎 + 𝐼 𝑏 + 𝐽𝑏 − 𝐺 𝑐 − 𝐻𝑐 ) +
𝑥(1 − 𝑦)(1 − 𝑧)(𝐷 − 𝐺 𝑎 − 𝐻𝑎 + 𝐼 𝑏 + 𝐽𝑏 + 𝐼 𝑐 + 𝐽𝑐 ) + (1 − 𝑥)(1 − 𝑦)(1 − 𝑧)(𝐷 + 𝐼 𝑎 + 𝐽𝑎 + 𝐼 𝑏 + 𝐽𝑏 + 𝐼 𝑐 + 𝐽𝑐 )
⇒ 𝐸𝑈𝑣𝐺 = 𝑥(−𝐺 𝑎 − 𝐻𝑎 − 𝐼𝑎 − 𝐽𝑎 ) + 𝑦(−𝐺 𝑏 − 𝐻𝑏 − 𝐼𝑏 − 𝐽𝑏 ) + 𝑧(−𝐺 𝑐 − 𝐻𝑐 − 𝐼𝑐 − 𝐽𝑐 ) + (𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐼𝑏 + 𝐽𝑏
+ 𝐼𝑐 + 𝐽𝑐 )

The expected utility of Subnational environmental governance choosing the 'Lax Regulation” strategy is:
𝐺
𝑇1−𝑣 = 𝑥𝑦𝑧(−𝐷) + (1 − 𝑥)𝑦𝑧(−𝐷 − 𝐸 𝑎 ) + 𝑥𝑦(1 − 𝑧)(−𝐷 − 𝐸 𝑐 ) + (1 − 𝑥)𝑦(1 − 𝑧)(−𝐷 − 𝐸 𝑎 − 𝐸 𝑐 ) +
𝑥(1 − 𝑦)𝑧(−𝐷 − 𝐸 𝑏 ) + (1 − 𝑥)(1 − 𝑦)𝑧(−𝐷 − 𝐸 𝑎 − 𝐸 𝑏 ) + 𝑥(1 − 𝑦)(1 − 𝑧)(−𝐷 − 𝐸 𝑏 − 𝐸 𝑐 ) +
(1 − 𝑥)(1 − 𝑦)(1 − 𝑧)(−𝐷 − 𝐸 𝑎 − 𝐸 𝑏 − 𝐸 𝑐 )

𝐺
⇒ 𝐸𝑈1−𝑣 = 𝑥(𝐸 𝑎 ) + 𝑦(𝐸 𝑏 ) + 𝑧(𝐸 𝑐 ) + (−𝐷 − 𝐸 𝑎 − 𝐸 𝑏 − 𝐸 𝑐 )

The expected utility of Subnational environmental governance is:

𝐸𝑈 𝐺 = 𝑣𝐸𝑈𝑣𝐺 + (1 − 𝑣)𝐸𝑈1−𝑣
𝐺

Based on each player's expected utility, the paper adopts replicator dynamics equations from EGT to evaluate
evolutionary stability.
The replicator dynamics equation of the strategy ‘ Environmental Regulation’ (v) of Subnational environmental
governance is:
𝑑𝑣
= 𝑣(𝐸𝑈𝑣𝐺 − 𝐸𝑈 𝐺 ) = 𝑣[𝐸𝑈𝑣𝐺 − 𝑣𝐸𝑈𝑣𝐺 − (1 − 𝑣)𝐸𝑈1−𝑣
𝐺 𝐺
] = 𝑣(1 − 𝑣)(𝐸𝑈𝑣𝐺 − 𝐸𝑈1−𝑣 )
𝑑𝑡

𝑑𝑣
⇒ = 𝑣(1 − 𝑣)[𝑥(−𝐺𝑎 − 𝐻𝑎 − 𝐼𝑎 − 𝐽𝑎 − 𝐸𝑎 ) + 𝑦(−𝐺𝑏 − 𝐻𝑏 − 𝐼𝑏 − 𝐽𝑏 − 𝐸𝑏 ) + 𝑧(−𝐺𝑐 − 𝐻𝑐 − 𝐼𝑐 − 𝐽𝑐 − 𝐸𝑐 ) +
𝑑𝑡
𝑎 𝑏 𝑐
(2𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐼𝑏 + 𝐽𝑏 + 𝐼𝑐 + 𝐽𝑐 + 𝐸 + 𝐸 + 𝐸 )]

b, Fruit orchards
The expected utility of Fruit orchards choosing the 'Resource Sharing and Waste Management” strategy is:
𝐸𝑈𝑥𝑎 = 𝑣𝑦𝑧(𝐷𝑎 + 𝐺 𝑎 + 𝐻 𝑎 + 𝐾𝑐𝑎 + 𝐿𝑎𝑐 + 𝑂𝑐𝑎 + 𝑀𝑏𝑎 ) + 𝑣𝑦(1 − 𝑧)(𝐷𝑎 + 𝐺 𝑎 + 𝐻𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 + 𝑀𝑏𝑎 ) +
𝑣(1 − 𝑦)𝑧(𝐷𝑎 + 𝐺 𝑎 + 𝐻𝑎 + 𝐾𝑐𝑎 + 𝐿𝑎𝑐 + 𝑂𝑐𝑎 − 𝑀𝑏𝑎 ) + 𝑣(1 − 𝑦)(1 − 𝑧)(𝐷 𝑎 + 𝐺 𝑎 + 𝐻𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 ) +
(1 − 𝑣)𝑦𝑧(𝐷 𝑎 − 𝐹 𝑎 + 𝐾𝑐𝑎 + 𝐿𝑎𝑐 + 𝑂𝑐𝑎 + 𝑀𝑏𝑎 ) + (1 − 𝑣)𝑦(1 − 𝑧)(𝐷𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 + 𝑀𝑏𝑎 ) +
(1 − 𝑣)(1 − 𝑦)𝑧(𝐷 𝑎 − 𝐹 𝑎 + 𝐾𝑐𝑎 + 𝐿𝑎𝑐 + 𝑂𝑐𝑎 − 𝑀𝑏𝑎 ) + (1 − 𝑣)(1 − 𝑦)(1 − 𝑧)(𝐷𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 )

⇒ 𝐸𝑈𝑥𝑎 = 𝑣(𝐺 𝑎 + 𝐻𝑎 + 𝐹 𝑎 ) + 𝑦(2𝑀𝑏𝑎 ) + 𝑧(2𝐾𝑐𝑎 + 2𝐿𝑎𝑐 + 2𝑂𝑐𝑎 ) + (𝐷𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 )

The expected utility of Fruit orchards choosing the 'No Resource Sharing and Waste Management” strategy is:
𝑎
𝐸𝑈1−𝑥 = 𝑣𝑦𝑧(−𝐷𝑎 − 𝐼 𝑎 − 𝐽𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 ) + 𝑣𝑦(1 − 𝑧)(−𝐷𝑎 − 𝐼 𝑎 − 𝐽𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 ) +
𝑣(1 − 𝑦)𝑧(−𝐷𝑎 − 𝐼 𝑎 − 𝐽𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 ) + 𝑣(1 − 𝑦)(1 − 𝑧)(−𝐷𝑎 − 𝐼 𝑎 − 𝐽𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 ) +
 (1 − 𝑣)𝑦𝑧(−𝐷 𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 ) + (1 − 𝑣)𝑦(1 − 𝑧)(−𝐷𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 ) +
(1 − 𝑣)(1 − 𝑦)𝑧(−𝐷 𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 ) + (1 − 𝑣)(1 − 𝑦)(1 − 𝑧)(−𝐷𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 )

𝑎
⇒ 𝐸𝑈1−𝑥 = (𝐹 𝑎 − 𝐼 𝑎 − 𝐽𝑎 )𝑣 + (−𝐷𝑎 − 𝐹 𝑎 − 𝐾𝑐𝑎 − 𝐿𝑎𝑐 − 𝑂𝑐𝑎 − 𝑀𝑏𝑎 )

The expected utility of Fruit orchards is:

𝐸𝑈 𝑎 = 𝑥𝐸𝑈𝑥𝑎 + (1 − 𝑥)𝐸𝑈1−𝑥
𝑎

The replicator dynamics equation of the strategy ‘Resource Sharing and Waste Management’ (x) of Fruit orchards is:
𝑑𝑥
= 𝑥(𝐸𝑈𝑥𝑎 − 𝐸𝑈 𝑎 ) = 𝑥[𝐸𝑈𝑥𝑎 − 𝑥𝐸𝑈𝑥𝑎 − (1 − 𝑥)𝐸𝑈1−𝑥
𝑎 𝑎
] = 𝑥(1 − 𝑥)(𝐸𝑈𝑥𝑎 − 𝐸𝑈1−𝑥 )
𝑑𝑡

𝑑𝑥
⇒ = 𝑥(1 − 𝑥)[𝑣(𝐺𝑎 + 𝐻𝑎 + 𝐼𝑎 + 𝐽𝑎 ) + 𝑦(2𝑀𝑎𝑏 ) + 𝑧(2𝐾𝑎𝑐 + 2𝐿𝑎𝑐 + 2𝑂𝑎𝑐 ) + 2𝐷𝑎 ]
𝑑𝑡

c, Bee apiaries
The expected utility of Bee apiaries choosing the 'Resource Sharing and Waste Management” strategy is:
𝐸𝑈𝑦𝑏 = 𝑣𝑥𝑧(𝐷 𝑏 + 𝐺 𝑏 + 𝐻𝑏 + 𝐾𝑎𝑏 + 𝐿𝑏𝑐 ) + 𝑣(1 − 𝑥)𝑧(𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 + 𝐿𝑏𝑐 ) +
𝑣𝑥(1 − 𝑧)(𝐷𝑏 + 𝐺 𝑏 + 𝐻 𝑏 + 𝐾𝑎𝑏 − 𝐿𝑏𝑐 ) + 𝑣(1 − 𝑥)(1 − 𝑧)(𝐷𝑏 + 𝐺 𝑏 + 𝐻𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 ) +
(1 − 𝑣)𝑥𝑧(𝐷𝑏 − 𝐹 𝑏 + 𝐾𝑎𝑏 + 𝐿𝑏𝑐 ) + (1 − 𝑣)(1 − 𝑥)𝑧(𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 + 𝐿𝑏𝑐 ) +c
(1 − 𝑣)𝑥(1 − 𝑧)(𝐷𝑏 − 𝐹 𝑏 + 𝐾𝑎𝑏 − 𝐿𝑏𝑐 ) + (1 − 𝑣)(1 − 𝑥)(1 − 𝑧)(𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 )

⇒ 𝐸𝑈𝑦𝑏 = 𝑣(𝐺 𝑏 + 𝐻𝑏 + 𝐹 𝑏 + 𝐾𝑎𝑏 ) + 𝑥(2𝐾𝑎𝑏 ) + 𝑧(2𝐿𝑏𝑐 ) + (𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 )

The expected utility of Bee apiaries choosing the 'No Resource Sharing and Waste Management” strategy is:
𝑏
𝐸𝑈1−𝑦 = 𝑣𝑥𝑧(−𝐷𝑏 − 𝐼 𝑏 − 𝐽𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 ) + 𝑣(1 − 𝑥)𝑧(−𝐷𝑏 − 𝐼 𝑏 − 𝐽𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 ) +
𝑣𝑥(1 − 𝑧)(−𝐷𝑏 − 𝐼 𝑏 − 𝐽𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 ) + 𝑣(1 − 𝑥)(1 − 𝑧)(−𝐷𝑏 − 𝐼 𝑏 − 𝐽𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 ) +
(1 − 𝑣)𝑥𝑧(−𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 ) + (1 − 𝑣)(1 − 𝑥)𝑧(−𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 ) +
(1 − 𝑣)𝑥(1 − 𝑧)(−𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 ) + (1 − 𝑣)(1 − 𝑥)(1 − 𝑧)(−𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 )

𝑏
⇒ 𝐸𝑈1−𝑦 = (𝐹 𝑏 − 𝐼 𝑏 − 𝐽𝑏 )𝑣 + (−𝐷𝑏 − 𝐹 𝑏 − 𝐾𝑎𝑏 − 𝐿𝑏𝑐 )

The expected utility of Bee apiaries is:

𝐸𝑈 𝑏 = 𝑦𝐸𝑈𝑦𝑏 + (1 − 𝑦)𝐸𝑈1−𝑦
𝑏

The replicator dynamics equation of the strategy ‘Resource Sharing and Waste Management’ (y) of Bee apiaries is:
𝑑𝑦
= 𝑦(𝐸𝑈𝑦𝑏 − 𝐸𝑈 𝑏 ) = 𝑦[𝐸𝑈𝑦𝑏 − 𝑦𝐸𝑈𝑦𝑏 − (1 − 𝑦)𝐸𝑈1−𝑦
𝑏 𝑏
] = 𝑦(1 − 𝑦)(𝐸𝑈𝑦𝑏 − 𝐸𝑈1−𝑦 )
𝑑𝑡

𝑑𝑦
⇒ = 𝑦(1 − 𝑦)[𝑣(𝐺𝑏 + 𝐻𝑏 + 𝐼𝑏 + 𝐽𝑏 ) + 𝑥(2𝐾𝑏𝑎 ) + 𝑧(2𝐿𝑏𝑐 ) + 2𝐷𝑏 ]
𝑑𝑡
d, Coffee farms
The expected utility of Coffee farms choosing the 'Resource Sharing and Waste Management” strategy is:
𝐸𝑈𝑧𝑐 = 𝑣𝑥𝑦(𝐷𝑐 + 𝐺 𝑐 + 𝐻𝑐 + 𝐾𝑎𝑐 + 𝐿𝑐𝑎 + 𝑂𝑎𝑐 + 𝑀𝑏𝑐 ) + 𝑣(1 − 𝑥)𝑦(𝐷 𝑐 + 𝐺 𝑐 + 𝐻𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 + 𝑀𝑏𝑐 ) +
𝑣𝑥(1 − 𝑦)(𝐷 𝑐 + 𝐺 𝑐 + 𝐻𝑐 + 𝐾𝑎𝑐 + 𝐿𝑐𝑎 + 𝑂𝑎𝑐 − 𝑀𝑏𝑐 ) + 𝑣(1 − 𝑥)(1 − 𝑦)(𝐷 𝑐 + 𝐺 𝑐 + 𝐻𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 ) +
(1 − 𝑣)𝑥𝑦(𝐷𝑐 − 𝐹 𝑐 + 𝐾𝑎𝑐 + 𝐿𝑐𝑎 + 𝑂𝑎𝑐 + 𝑀𝑏𝑐 ) + (1 − 𝑣)(1 − 𝑥)𝑦(𝐷 𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 + 𝑀𝑏𝑐 ) +
(1 − 𝑣)𝑥(1 − 𝑦)(𝐷𝑐 − 𝐹 𝑐 + 𝐾𝑎𝑐 + 𝐿𝑐𝑎 + 𝑂𝑎𝑐 − 𝑀𝑏𝑐 ) + (1 − 𝑣)(1 − 𝑥)(1 − 𝑦)(𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 )

⇒ 𝐸𝑈𝑧𝑐 = 𝑣(𝐺 𝑐 + 𝐻𝑐 + 𝐹 𝑐 ) + 𝑥(2𝐾𝑎𝑐 + 2𝐿𝑐𝑎 + 2𝑂𝑎𝑐 ) + 𝑦(2𝑀𝑏𝑐 ) + (𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 )

The expected utility of Coffee farms choosing the 'No Resource Sharing and Waste Management” strategy is:
𝑐
𝐸𝑈1−𝑧 = 𝑣𝑥𝑦(−𝐷 𝑐 − 𝐼 𝑐 − 𝐽𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 ) + 𝑣(1 − 𝑥)(1 − 𝑧)(−𝐷𝑐 − 𝐼 𝑐 − 𝐽𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 ) +
𝑣𝑥(1 − 𝑦)(−𝐷 𝑐 − 𝐼 𝑐 − 𝐽𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 ) + 𝑣(1 − 𝑥)(1 − 𝑦)(−𝐷𝑐 − 𝐼 𝑐 − 𝐽𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 ) +
(1 − 𝑣)𝑥𝑦(−𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 ) + (1 − 𝑣)(1 − 𝑥)𝑦(−𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 ) +
(1 − 𝑣)𝑥(1 − 𝑦)(−𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 ) + (1 − 𝑣)(1 − 𝑥)(1 − 𝑦)(−𝐷𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 )

𝑐
⇒ 𝐸𝑈1−𝑧 = (𝐹 𝑐 − 𝐼 𝑐 − 𝐽𝑐 )𝑣 + (−𝐷 𝑐 − 𝐹 𝑐 − 𝐾𝑎𝑐 − 𝐿𝑐𝑎 − 𝑂𝑎𝑐 − 𝑀𝑏𝑐 )

The expected utility of Coffee farms is:

𝐸𝑈 𝑐 = 𝑧𝐸𝑈𝑧𝑐 + (1 − 𝑧)𝐸𝑈1−𝑧
𝑐

The replicator dynamics equation of the strategy ‘Resource Sharing and Waste Management’ (z) of Coffee farms is:
𝑑𝑧
= 𝑧(𝐸𝑈𝑧𝑐 − 𝐸𝑈 𝑐 ) = 𝑧[𝐸𝑈𝑧𝑐 − 𝑧𝐸𝑈𝑧𝑐 − (1 − 𝑧)𝐸𝑈1−𝑧
𝑐 𝑐
] = 𝑧(1 − 𝑧)(𝐸𝑈𝑧𝑐 − 𝐸𝑈1−𝑧 )
𝑑𝑡

𝑑𝑧
⇒ = 𝑧(1 − 𝑧)[𝑣(𝐺𝑐 + 𝐻𝑐 + 𝐼𝑐 + 𝐽𝑐 ) + 𝑥(2𝐾𝑐𝑎 + 2𝐿𝑐𝑎 + 2𝑂𝑐𝑎 ) + 𝑦(2𝑀𝑐𝑏 ) + 2𝐷𝑐 ]
𝑑𝑡

6. Stability analysis of strategy combination

According to the method proposed by Denial Friedman, the ESS of a differential equation system can be obtained
from the local stability analysis of the Jacobian matrix of the system.

A system of replicator dynamics equations is constructed as follows:

𝑑𝑣
= 𝑣(1 − 𝑣)[𝑥(−𝐺 𝑎 − 𝐻𝑎 − 𝐼 𝑎 − 𝐽𝑎 − 𝐸 𝑎 ) + 𝑦(−𝐺 𝑏 − 𝐻 𝑏 − 𝐼 𝑏 − 𝐽𝑏 − 𝐸 𝑏 ) + 𝑧(−𝐺 𝑐 − 𝐻𝑐 − 𝐼 𝑐 − 𝐽𝑐 − 𝐸 𝑐 ) +
𝑑𝑡
 (2𝐷 + 𝐼 𝑎 + 𝐽𝑎 + 𝐼 𝑏 + 𝐽𝑏 + 𝐼 𝑐 + 𝐽𝑐 + 𝐸 𝑎 + 𝐸 𝑏 + 𝐸 𝑐 )]

𝑑𝑥
= 𝑥(1 − 𝑥)[𝑣(𝐺 𝑎 + 𝐻𝑎 + 𝐼 𝑎 + 𝐽𝑎 ) + 𝑦(2𝑀𝑏𝑎 ) + 𝑧(2𝐾𝑐𝑎 + 2𝐿𝑎𝑐 + 2𝑂𝑐𝑎 ) + 2𝐷𝑎 ]
𝑑𝑡

𝑑𝑦
= 𝑦(1 − 𝑦)[𝑣(𝐺 𝑏 + 𝐻𝑏 + 𝐼 𝑏 + 𝐽𝑏 ) + 𝑥(2𝐾𝑎𝑏 ) + 𝑧(2𝐿𝑏𝑐 ) + 2𝐷𝑏 ]
𝑑𝑡

𝑑𝑧
= 𝑧(1 − 𝑧)[𝑣(𝐺 𝑐 + 𝐻𝑐 + 𝐼 𝑐 + 𝐽𝑐 ) + 𝑥(2𝐾𝑎𝑐 + 2𝐿𝑐𝑎 + 2𝑂𝑎𝑐 ) + 𝑦(2𝑀𝑏𝑐 ) + 2𝐷𝑐 ]
𝑑𝑡

According to EGT, a strategy reaches a stable equilibrium when its corresponding replicator dynamics equation
equals zero. For the Subnational Environmental Governance to consistently adopt the “Environmental Regulation”
strategy and for Fruit Orchards, Bee Apiaries, and Coffee Farms to select the “Resource Sharing and Waste
Management” strategy, these choices must qualify as Evolutionarily Stable Strategies (ESS). This condition ensures
that, once established, these strategies cannot be invaded or displaced by alternative strategies within the population
dynamics.

𝑑𝑣 𝑑𝑥 𝑑𝑦 𝑑𝑧
Which mean = = = = 0. Solving the replicator dynamics equations yields a total of 16 equilibrium
𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡

points. These equilibria may be located either at the fixed points within the interior of the strategy space or along its
boundaries. Not all equilibrium points correspond to Evolutionarily Stable Strategies (ESS); those situated in the
interior may serve as centroids or saddle points, indicating dynamic neutrality or instability rather than evolutionary
robustness. Only those equilibria that satisfy the strict stability conditions of ESS can be considered resilient against
strategic deviations in the evolutionary game framework. Accordingly, the asymptotic stability of the equilibrium
points is examined by analyzing the local stability properties of the Jacobian matrix evaluated at each equilibrium
𝑑𝑣 𝑑𝑥 𝑑𝑦 𝑑𝑧
based on method proposed by Daneil Friedman [124]. By computing the partial derivatives of , , , with
𝑑𝑡 𝑑𝑡 𝑑𝑡 𝑑𝑡

respect to 𝑣, 𝑥, 𝑦, 𝑧, the Jacobian matrix of the replication dynamic system can be obtained as:
𝑎 𝑏 𝑐
𝛼11 = (1 − 2𝑣)[𝑥(−𝐺 𝑎 − 𝐻 𝑎 − 𝐼𝑎 − 𝐽𝑎 − 𝐸 ) + 𝑦(−𝐺 𝑏 − 𝐻 𝑏 − 𝐼𝑏 − 𝐽𝑏 − 𝐸 ) + 𝑧(−𝐺 𝑐 − 𝐻 𝑐 − 𝐼𝑐 − 𝐽𝑐 − 𝐸 ) +
𝑎 𝑏 𝑐
(2𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐼𝑏 + 𝐽𝑏 + 𝐼𝑐 + 𝐽𝑐 + 𝐸 + 𝐸 + 𝐸 )]
𝑎
𝛼12 = 𝑣(1 − 𝑣)(−𝐺 𝑎 − 𝐻 𝑎 − 𝐼𝑎 − 𝐽𝑎 − 𝐸 )
𝑏
𝛼13 = 𝑣(1 − 𝑣)(−𝐺 𝑏 − 𝐻 𝑏 − 𝐼𝑏 − 𝐽𝑏 − 𝐸 )
𝑐
𝛼14 = 𝑣(1 − 𝑣)(−𝐺 𝑐 − 𝐻 𝑐 − 𝐼𝑐 − 𝐽𝑐 − 𝐸 )

𝑎 𝑎 𝑎 𝑎
𝛼21 = 𝑥(1 − 𝑥)(𝐺 + 𝐻 + 𝐼 + 𝐽 )
𝑎 𝑎 𝑎 𝑎 𝑎 𝑎 𝑎 𝑎 𝑎
𝛼22 = (1 − 2𝑥)[𝑣(𝐺 + 𝐻 + 𝐼 + 𝐽 ) + 𝑦(2𝑀𝑏 ) + 𝑧(2𝐾𝑐 + 2𝐿𝑐 + 2𝑂𝑐 ) + 2𝐷 ]
𝑎
𝛼23 = 𝑥(1 − 𝑥)2𝑀𝑏
𝑎 𝑎 𝑎
𝛼24 = 𝑥(1 − 𝑥)(2𝐾𝑐 + 2𝐿𝑐 + 2𝑂𝑐 )
 𝑏 𝑏 𝑏 𝑏
𝛼31 = 𝑦(1 − 𝑦)(𝐺 + 𝐻 + 𝐼 + 𝐽 )
𝑏
𝛼32 = 𝑦(1 − 𝑦)2𝐾𝑎
𝑏 𝑏 𝑏 𝑏 𝑏 𝑏 𝑏
𝛼33 = (1 − 2𝑦)[𝑣(𝐺 + 𝐻 + 𝐼 + 𝐽 ) + 𝑥(2𝐾𝑎 ) + 𝑧(2𝐿𝑐 ) + 2𝐷 ]
𝑏
𝛼34 = 𝑦(1 − 𝑦)2𝐿𝑐

𝑐 𝑐 𝑐 𝑐
𝛼41 = 𝑧(1 − 𝑧)(𝐺 + 𝐻 + 𝐼 + 𝐽 )
𝑐
𝛼42 = + 2𝐿𝑐𝑎 + 2𝑂𝑐𝑎 )
𝑧(1 − 𝑧)(2𝐾𝑎
𝑐
𝛼43 = 𝑧(1 − 𝑧)2𝑀𝑏
𝛼44 = (1 − 2𝑧)[𝑣(𝐺𝑐 + 𝐻𝑐 + 𝐼𝑐 + 𝐽𝑐 ) + 𝑥(2𝐾𝑐𝑎 + 2𝐿𝑐𝑎 + 2𝑂𝑐𝑎 ) + 𝑦(2𝑀𝑐𝑏 ) + 2𝐷𝑐 ]

By substituting the 16 pure strategy equilibrium points into the Jacobian matrix, the corresponding eigenvalues for
each point are derived. A summary of these results is presented in Table 7.

Equilibrium Eigenvalues Stability

(v, x, y, z)

(1,0,0,0) 𝜆1 = −(2𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐼𝑏 + 𝐽𝑏 + 𝐼𝑐 + 𝐽𝑐 + 𝐸 + 𝐸 + 𝐸 ) < 0

𝑎 𝑏 𝑐 Unstable
𝑎 𝑎 𝑎 𝑎 𝑎
𝜆2 = 𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 > 0
𝑏 𝑏 𝑏 𝑏 𝑏
𝜆3 = 𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 > 0
𝑐 𝑐 𝑐 𝑐 𝑐
𝜆4 = 𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 > 0

(1,1,0,0) 𝜆1 = −(2𝐷 + 𝐼𝑏 + 𝐽𝑏 + 𝐼𝑐 + 𝐽𝑐 + 𝐸 + 𝐸 − 𝐺 𝑎 − 𝐻 𝑎 )
𝑏 𝑐 Unstable
𝑎 𝑎 𝑎 𝑎 𝑎
𝜆2 = −(𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 ) < 0
𝑏 𝑏 𝑏 𝑏 𝑏 𝑏
𝜆3 = 𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 + 2𝐾𝑎 > 0
𝑐 𝑐 𝑐 𝑐 𝑐 𝑐 𝑐 𝑐
𝜆4 = 𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 + 2𝐾𝑎 + 2𝐿𝑎 + 2𝑂𝑎 > 0

𝑎 𝑐
(1,0,1,0) 𝜆1 = −(2𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐼𝑐 + 𝐽𝑐 + 𝐸 + 𝐸 − 𝐺 𝑏 − 𝐻 𝑏 ) Unstable
𝑎 𝑎 𝑎 𝑎 𝑎
𝜆2 = 𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 + 2𝑀𝑎𝑏 >0
𝑏 𝑏 𝑏 𝑏 𝑏
𝜆3 = −(𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 ) < 0
𝑐 𝑐 𝑐 𝑐 𝑐 𝑐
𝜆4 = 𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 + 2𝑀𝑏 > 0

𝑎 𝑐
(1,0,0,1) 𝜆1 = −(2𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐼𝑏 + 𝐽𝑏 + 𝐸 + 𝐸 − 𝐺 𝑐 − 𝐻 𝑐 ) Unstable
𝑎 𝑎 𝑎 𝑎 𝑎 𝑎 𝑎 𝑎
𝜆2 = 𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 + 2𝐾𝑐 + 2𝐿𝑐 + 2𝑂𝑐 > 0
𝑏 𝑏 𝑏 𝑏 𝑏 𝑏
𝜆3 = 𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 + 2𝐿𝑐 > 0
𝑐 𝑐 𝑐 𝑐 𝑐
𝜆4 = −(𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 ) < 0

𝑐
(1,1,1,0) 𝜆1 = −(2𝐷 + 𝐼𝑐 + 𝐽𝑐 + 𝐸 − 𝐺 𝑎 − 𝐻 𝑎 − 𝐺 𝑏 − 𝐻 𝑏 ) Unstable
𝑎 𝑎 𝑎 𝑎 𝑎
𝜆2 = −(𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 + 2𝑀𝑎𝑏 ) <0
 𝑏 𝑏 𝑏 𝑏 𝑏 𝑏
𝜆3 = −(𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 + 2𝐾𝑎 ) < 0
𝑐
𝜆4 = 𝐺 + 𝐻 + 𝐼 + 𝐽
𝑐 𝑐 𝑐
+ 2𝐾𝑐𝑎 + 2𝐿𝑐𝑎 + 2𝑂𝑐𝑎 + 2𝑀𝑐𝑏 + 2𝐷𝑐 > 0

𝑎
(1,0,1,1) 𝜆1 = −(2𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐸 − 𝐺 𝑏 − 𝐻 𝑏 − 𝐺 𝑐 − 𝐻 𝑐 ) Unstable
𝑎
𝜆2 = 𝐺 + 𝐻 + 𝐼 + 𝐽
𝑎 𝑎 𝑎
+ 2𝑀𝑎𝑏 + 2𝐾𝑎𝑐 + 2𝐿𝑎𝑐 + 2𝑂𝑎𝑐 + 2𝐷 > 0 𝑎

𝑏 𝑏 𝑏 𝑏 𝑏 𝑏
𝜆3 = −(𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 + 2𝐿𝑐 ) <0
𝑐 𝑐 𝑐 𝑐 𝑐 𝑐
𝜆4 = −(𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 + 2𝑀𝑏 ) <0

(1,1,0,1) 𝜆1 = −(2𝐷 + 𝐼𝑏 + 𝐽𝑏 + 𝐸 − 𝐺 𝑎 − 𝐻 𝑎 − 𝐺 𝑐 − 𝐻 𝑐 )
𝑏 Unstable
𝑎 𝑎
𝜆2 = −(𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷
𝑎 𝑎 𝑎
+ 2𝐾𝑎𝑐 + 2𝐿𝑎𝑐 + 2𝑂𝑎𝑐 ) <0
𝑏 𝑏 𝑏 𝑏 𝑏 𝑏 𝑏
𝜆3 = 𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐾𝑎 + 2𝐿𝑐 + 2𝐷 > 0
𝑐 𝑐 𝑐 𝑐 𝑐 𝑐 𝑐 𝑐
𝜆4 = −(𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 + 2𝐾𝑎 + 2𝐿𝑎 + 2𝑂𝑎 ) < 0

(1,1,1,1) 𝜆1 = −(2𝐷 − 𝐺 𝑎 − 𝐻 𝑎 − 𝐺 𝑏 − 𝐻 𝑏 − 𝐺 𝑐 − 𝐻 𝑐 ) Stable if

𝑎 𝑎 𝑎 𝑎
+ 2𝑀𝑎𝑏 + 2𝐾𝑎𝑐 + 2𝐿𝑎𝑐 + 2𝑂𝑎𝑐 𝑎 𝐺 + 𝐻 + 𝐺 𝑏 + 𝐻𝑏 + 𝐺 𝑐 + 𝐻𝑐
𝑎 𝑎
𝜆2 = −(𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐷 ) < 0
𝑏 𝑏 𝑏 𝑏 𝑏 𝑏 𝑏 < 2𝐷
𝜆3 = −(𝐺 + 𝐻 + 𝐼 + 𝐽 + 2𝐾𝑎 + 2𝐿𝑐 + 2𝐷 ) < 0
𝑐
𝜆4 = −(𝐺 + 𝐻 + 𝐼 + 𝐽
𝑐 𝑐 𝑐
+ 2𝐾𝑐𝑎 + 2𝐿𝑐𝑎 + 2𝑂𝑐𝑎 + 2𝑀𝑐𝑏 + 2𝐷𝑐 < 0

Table 7: Eigenvalues of the four-player game equilibrium points

As shown in Table 7, the only asymptotically stable equilibrium occurs at the point (1,1,1,1), where the Subnational
Environmental Governance adopts the “Environmental Regulation” strategy, and the Fruit Orchards, Bee Apiaries,
and Coffee Farms all choose the “Resource Sharing and Waste Management” strategy.

The equilibrium is stable only when 𝐺 𝑎 + 𝐻𝑎 + 𝐺 𝑏 + 𝐻𝑏 + 𝐺 𝑐 + 𝐻𝑐 < 2𝐷 or (𝐺 𝑎 + 𝐺 𝑏 + 𝐺 𝑐 ) + (𝐻𝑎 + 𝐻𝑏 +

𝐻𝑐 ) < 𝐷 + 𝐷. This result suggests that the reputation value of the Subnational Environmental Governance exceeds
the total incentive payments required to motivate corporate actors to adopt either the “Resource Sharing” or “Waste
Management” strategies. From the perspective of the regulator, this is a rational outcome: reputation serves as a
foundational asset that facilitates broader policy effectiveness, including the ability to attract funding and enforce
compliance. In contrast, without a credible reputation, even substantial financial incentives may fail to secure
participation from stakeholders, undermining the long-term viability of cooperative environmental strategies.

Evaluation
For governments, reputation and public trust are foundational assets that outweigh the short-term accumulation of
financial capital. A strong reputation enhances a government's legitimacy, signaling fairness, competence, and
accountability, which in turn fosters public trust and voluntary compliance. This legitimacy is critical for maintaining
long-term social stability and minimizing the risk of unrest or governance breakdown. Beyond domestic
implications, a reputable government is more likely to be viewed as a reliable partner on the international stage,
opening doors to favorable diplomatic relations, foreign investment, and global trade. Economically, trust in
governance attracts sustainable investment, encourages entrepreneurship, and supports policy implementation, all of
which are vital for long-term growth. Moreover, a government that values its reputation is more inclined to act with
foresight, balancing immediate fiscal needs with broader social and environmental responsibilities. In contrast, while
monetary resources are essential for operational capacity, they are ultimately a means, not an end. Prioritizing
financial gain at the expense of trust and credibility can erode institutional integrity and weaken the social contract,
resulting in long-term governance challenges that no amount of funding alone can resolve.

Equilibrium Eigenvalues Stability

(v,x,y,z)

(0,0,0,0) 𝜆1 = 2𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐼𝑏 + 𝐽𝑏 + 𝐼𝑐 + 𝐽𝑐 + 𝐸 + 𝐸 + 𝐸 > 0
𝑎 𝑏 𝑐 Unstable
𝑎
𝜆2 = 2𝐷 > 0
𝑏
𝜆3 = 2𝐷 > 0
𝑐
𝜆4 = 2𝐷 > 0

(0,1,0,0) 𝜆1 = 2𝐷 + 𝐼𝑏 + 𝐽𝑏 + 𝐼𝑐 + 𝐽𝑐 + 𝐸 + 𝐸 − 𝐺 𝑎 − 𝐻 𝑎
𝑏 𝑐 Unstable
𝑎
𝜆2 = −(2𝐷 ) < 0
𝑏 𝑏
𝜆3 = 2𝐷 + 2𝐾𝑎 > 0
𝑐 𝑐 𝑐 𝑐
𝜆4 = 2𝐷 + 2𝐾𝑎 + 2𝐿𝑎 + 2𝑂𝑎 > 0

𝑎 𝑐
(0,0,1,0) 𝜆1 = 2𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐼𝑐 + 𝐽𝑐 + 𝐸 + 𝐸 − 𝐺 𝑏 − 𝐻 𝑏 Unstable
𝑎 𝑎
𝜆2 = 2𝐷 + 2𝑀𝑏 > 0
𝑏
𝜆3 = −(2𝐷 ) < 0
𝑐 𝑐
𝜆4 = 2𝐷 + 2𝑀𝑏 > 0

𝑎 𝑐
(0,0,0,1) 𝜆1 = 2𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐼𝑏 + 𝐽𝑏 + 𝐸 + 𝐸 − 𝐺 𝑐 − 𝐻 𝑐 Unstable
𝑎
𝜆2 = 2𝐷 + 2𝐾𝑎𝑐 + 2𝐿𝑎𝑐 + 2𝑂𝑎𝑐 >0
𝑏
𝜆3 = 2𝐷 + 2𝐿𝑏𝑐 >0
𝑐
𝜆4 = −(2𝐷 ) < 0

𝑐
(0,1,1,0) 𝜆1 = 2𝐷 + 𝐼𝑐 + 𝐽𝑐 + 𝐸 − 𝐺 𝑎 − 𝐻 𝑎 − 𝐺 𝑏 − 𝐻 𝑏 Unstable
𝑎
𝜆2 = −(2𝐷 + 2𝑀𝑎𝑏 ) <0
𝑏 𝑏
𝜆3 = −(2𝐷 + 2𝐾𝑎 ) < 0
𝑐 𝑐 𝑐 𝑐 𝑐
𝜆4 = 2𝐾𝑎 + 2𝐿𝑎 + 2𝑂𝑎 + 2𝑀𝑏 + 2𝐷 > 0

𝑎
(0,0,1,1) 𝜆1 = 2𝐷 + 𝐼𝑎 + 𝐽𝑎 + 𝐸 − 𝐺 𝑏 − 𝐻 𝑏 − 𝐺 𝑐 − 𝐻 𝑐 Unstable
𝜆2 = 2𝑀𝑎𝑏 + 2𝐾𝑎𝑐 + 2𝐿𝑎𝑐 + 2𝑂𝑎𝑐 + 2𝐷 > 0 𝑎

𝑏 𝑏
𝜆3 = −(2𝐷 + 2𝐿𝑐 ) < 0
𝑐 𝑐
𝜆4 = −(2𝐷 + 2𝑀𝑏 ) < 0
 (0,1,0,1) 𝑏
𝜆1 = 2𝐷 + 𝐼𝑏 + 𝐽𝑏 + 𝐸 − 𝐺 𝑎 − 𝐻 𝑎 − 𝐺 𝑐 − 𝐻 𝑐 Unstable
𝑎
𝜆2 = −(2𝐷 + 2𝐾𝑎𝑐 + 2𝐿𝑎𝑐 + 2𝑂𝑎𝑐 ) <0
𝜆3 = 2𝐾𝑏𝑎 + 2𝐿𝑏𝑐 + 2𝐷𝑏 > 0
𝜆4 = −(2𝐷
𝑐
+ 2𝐾𝑐𝑎 + 2𝐿𝑐𝑎 + 2𝑂𝑐𝑎 ) <0

(0,1,1,1) 𝜆1 = 2𝐷 − 𝐺 𝑎 − 𝐻 𝑎 − 𝐺 𝑏 − 𝐻 𝑏 − 𝐺 𝑐 − 𝐻 𝑐 Stable if
𝑎
+ 2𝐾𝑎𝑐 + 2𝐿𝑎𝑐 + 2𝑂𝑎𝑐 𝑎 2𝐷 < 𝐺 + 𝐻 𝑎 + 𝐺 𝑏 + 𝐻 𝑏 + 𝐺 𝑐
𝑎
𝜆2 = −(2𝑀𝑏 + 2𝐷 ) < 0
𝑏 𝑏 𝑏 + 𝐻𝑐
𝜆3 = −(2𝐾𝑎 + 2𝐿𝑐 + 2𝐷 ) < 0
𝑐 𝑐 𝑐 𝑐 𝑐
𝜆4 = −(2𝐾𝑎 + 2𝐿𝑎 + 2𝑂𝑎 + 2𝑀𝑏 + 2𝐷 < 0

Table 8: Eigenvalues of the four-player game equilibrium points

As shown in Table 8, the only asymptotically stable equilibrium occurs at the point (0,1,1,1), where the Subnational
Environmental Governance adopts the “Lax Regulation” strategy, while the Fruit Orchards, Bee Apiaries, and Coffee
Farms continue to choose the “Resource Sharing and Waste Management” strategy.

The condition for equilibrium is 2𝐷 < 𝐺 𝑎 + 𝐻𝑎 + 𝐺 𝑏 + 𝐻𝑏 + 𝐺 𝑐 + 𝐻𝑐 or 𝐷 + 𝐷 < (𝐺 𝑎 + 𝐺 𝑏 + 𝐺 𝑐 ) + (𝐻 𝑎 + 𝐻𝑏 +

𝐻𝑐 ). This outcome indicates that the perceived value of the Subnational Environmental Governance’s reputation is
lower than the total incentive payments required to induce corporate actors to adopt either the “Resource Sharing” or
“Waste Management” strategies. Such a scenario is plausible when the government faces liquidity constraints or
carries a high level of public debt. Under these conditions, the regulator may lack both the financial capacity and the
political will to allocate resources toward incentive programs, resulting in a preference for lax regulation despite the
continued cooperation of corporate stakeholders.

In conclusion, it is both reasonable and optimal for the Subnational Environmental Governance to adopt the
“Environmental Regulation” strategy and for Fruit Orchards, Bee Apiaries, and Coffee Farms to engage in “Resource
Sharing and Waste Management” when the government places a higher value on institutional reputation than on
incentive expenditures, or when fiscal conditions allow for manageable debt levels.

7. Simulation

The paper provided a parameter table based on previous results from foundation and similar models. There is no
similar model including all SAFE players, however there are successful models with actual data within two players
from the SAFE model.

Subnational governance Fruit orchards Bee apiaries Coffee farms

𝐷 = 0.5 𝐷 𝑎 = 0.5 𝐷 𝑏 = 0.5 𝐷 𝑐 = 0.5

 𝐸 = 0.2 𝐾𝑐𝑎 = 0.5 𝐾𝑎𝑏 = 0.4 𝐾𝑎𝑐 = 0.6

𝐹 = 0.3 𝐿𝑎𝑐 = 0.3 𝐿𝑏𝑐 = 0.3 𝐿𝑐𝑎 = 0.5

𝐺𝑎 = 𝐺𝑏 = 𝐺𝑐 = 0.4 𝑂𝑐𝑎 = 0.2 𝑂𝑎𝑐 = 0.4

𝐻𝑎 = 𝐻𝑏 = 𝐻𝑐 = 0.6 𝑀𝑏𝑎 = 0.2 𝑀𝑏𝑐 = 0.5

𝐼𝑎 = 𝐼𝑏 = 𝐼𝑐 = 0.3

𝐽𝑎 = 𝐽𝑏 = 𝐽𝑐 = 0.4

Table 9: Parameter based on actual publicity data from news and privacy data from survey [40]; [93]; [12]; [55]; [4];
[60]; [73]; [46]; [92]; [97]; [41]; [87]; [78]; [86]; [3]; [56]; [91]; [96]; [45]; [57]; [72]; [51]; [7]; [59]; [29]; [63]

All data derive from publicly available farm records documenting integrated agricultural models, including bee-
coffee and durian-bee systems which demonstrated enhanced profitability following implementation. Additionally,
the study incorporates survey data from farms in Dak Lak Province. Given that income increases vary across farms
with differing models and scales, the analysis standardizes results by calculating average percentage increase in
income on a uniform scale. For unquantifiable factors such as reputation, a neutral assumption is applied. Waste
management is prioritized over resource sharing due to its critical environmental implications, warranting stronger
incentives and sanctions, a focus aligned with government policy priorities [38].

Figure 6: Time Plot

The simulation results reveal a highly cooperative equilibrium across all agricultural stakeholders under the specified
parameterization, with Subnational Environmental Governance exhibiting nuanced strategic stability while Fruit
Orchards, Bee Apiaries, and Coffee Farms demonstrate robust convergence to full cooperation. Subnational
Governance (v) witnessed an unstabilized strategy chosen across diverse initial conditions. This reflects the
governance role as a mediator balancing policy incentives against administrative constraints, with trajectories from
high-initial-cooperation states plateauing rapidly, while intermediate initial conditions exhibit delayed convergence,
suggesting adaptive regulatory learning.

In contrast, Fruit Orchards (x), Bee Apiaries (y), and Coffee Farms (z) converge monotonically to 𝑥 = 𝑦 = 𝑧 = 1
driven by the dominance of mutualistic economic incentives: intercropping benefits, waste management rewards and
pollination synergies overwhelm defection penalties

The near-synchronous convergence of y and z trajectories underscores co-evolutionary reinforcement, where nectar-
pollination dependencies and intercropping efficiencies create self-reinforcing cooperative feedback loops. Critically,
this system exhibits structural resilience, no bifurcations or oscillations occur, indicating that the current incentive
structure has created a globally stable cooperative attractor. This aligns with empirical observations in Vietnamese
integrated farming systems, where the SAFE model symbioses yield rapid adoption due to tangible economic gains,
validating the model’s core premise that mutualistic economic incentives, rather than regulatory coercion, are the
primary drivers of sustainable cooperation. The findings carry direct policy implications: optimizing reputational
benefits and intercropping incentives could further strengthen cooperation. This represents a critical advance over
traditional regulatory models, demonstrating that economic-mutualistic frameworks rather than punitive governance
sustain long-term environmental outcomes in agricultural networks.

Figure 7: Phase Portrait Coffee-Fruit

The model's synchronized convergence of Fruit Orchards and Coffee Farms directly mirrors the empirically
documented mutualistic synergy in Vietnamese agro-ecosystems, where coffee-fruit intercropping systems such as
those prevalent in Dak Lak province generate tangible economic and ecological co-benefits that drive rapid adoption.
As evidenced by field studies reporting around $2 thousand annual profit increases from integrated farming, the
model's parameters quantitatively capture these real-world dynamics. This interdependence manifests through three
mutually reinforcing mechanisms: (1) ecological complementarity (fruit trees provide shade and soil stabilization for
coffee, reducing irrigation costs), (2) resource efficiency (coffee residues enrich soil fertility for fruit crops), and (3)
market resilience (diversified income against price volatility). Critically, the model's near-identical convergence
trajectories for both groups, regardless of initial conditions, validate that these synergies create a self-reinforcing
cooperative attractor, aligning with empirical observations that farmers adopt intercropping within 1-2 seasons when
economic incentives exceed transaction costs. This underscores a pivotal policy insight: rather than relying on
regulatory coercion, interventions should amplify these existing mutualistic pathways such as subsidies for
intercropping infrastructure or pollination services to harness the system's inherent stability, as demonstrated by the
robustness of the 𝑥 = 𝑧 = 1 equilibrium across all simulated scenarios.

Figure 8: Phase Portrait Coffee-Bee

The phase portrait of Bee Apiaries versus Coffee Farms reveals a tightly coupled cooperative dynamic that directly
mirrors the mutually reinforcing ecological and economic relationships observed in Vietnamese coffee-bee agro-
ecosystems. All trajectories regardless of initial conditions converge to the full cooperation equilibrium reflecting the
real-world interdependence where coffee plantations provide critical nectar resources for bees while bee pollination
boosts coffee yields by 20%, as documented in field studies from Dak Lak province. This convergence aligns
precisely with empirical adoption patterns: farmers rapidly shift to integrated systems when economic gains
outweigh transaction costs, typically within 1 to 2 growing seasons. Critically, the model captures the self-
reinforcing nature of this relationship: coffee’s flowering cycles sustain bee populations, while bees’ pollination
services directly increase coffee productivity, creating a feedback loop that stabilizes cooperation. This dynamic
validates the model’s parameterization against real-world data, where integrated bee-coffee farming has generated
$1.5 thousand higher annual profits per hectare compared to monoculture systems. The absence of alternative
equilibria underscores that the current incentive structure has created an irreversibly stable cooperative attractor,
making regulatory intervention unnecessary. Instead, policymakers should prioritize amplifying these natural
synergies through targeted beekeeping subsidies and crop diversification programs.
 Figure 9: Phase Portrait Fruit-Bee

The phase portrait of Bee Apiaries versus Fruit Orchards illustrates a strongly synergistic dynamic rooted in their
real-world ecological and economic interdependence. All trajectories converge to the full cooperation equilibrium
regardless of initial conditions, reflecting the mutualistic relationship where Fruit Orchards provide nectar resources
for Bee Apiaries and Bee Apiaries enhance fruit pollination. This mirrors empirical observations in Vietnamese agro-
systems where integrated fruit-bee systems report yield increases of 50% due to improved pollination. Notably, the
absence of alternative equilibria underscores the structural stability of the cooperative attractor, validated by real-
world resilience: once established, these systems rarely revert to non-cooperation. Policymakers should prioritize
interventions amplifying these synergies, such as promoting pollinator-friendly fruit varieties or subsidizing
beekeeping infrastructure as demonstrated in successful programs in Dak Lak province. This analysis reaffirms that
evolutionary game models, when calibrated with granular mutualistic parameters, can accurately predict real-world
sustainable outcomes.
 Figure 10: Bifurcation of reputation benefit impact on strategy chosen

When subnational environmental governance entities receive higher reputational benefits (D) for enforcing
environmental regulations, they are incentivized to adopt stricter regulatory practices over time. This aligns with
real-life scenarios where agencies gain public trust, political credibility, or institutional recognition for proactive
compliance, motivating them to maintain rigorous standards. As these subnational actors prioritize strict
enforcement, the central government observing their reliability may feel empowered to implement even stricter
regulations at the national level. For instance, if a state agency successfully enforces carbon emission limits and
gains acclaim for its transparency, the federal government might raise national targets, knowing local bodies can
effectively implement them. Conversely, lower reputational benefits (D) might lead subnational entities to relax
enforcement (to avoid costs or conflicts), prompting the central government to either impose tighter controls directly
or face weaker overall compliance. Thus, higher D creates a virtuous cycle: strong local enforcement fosters national
confidence, enabling stricter policies that rely on trusted subnational partners.

The paper illustrated 3 different types of graphs including Time plot, Phase Portrait and Bifurcation. The trio
diagrams are essential for a comprehensive analysis of evolutionary dynamics, each fulfilling a distinct academic
function that collectively validates the model’s theoretical and policy contributions. Time series plots capture the
dynamic trajectory of strategy adoption over time, revealing how mutualistic incentives drive rapid convergence to
full cooperation directly mirroring Vietnamese farmers’ real-world adoption patterns. Phase portraits visualize the
basins of attraction in strategy space, proving the global stability of the cooperative equilibrium where all 50
trajectories from diverse initial conditions converge for Resource sharing and Waste management for Fruit Orchards,
Bee apiaries and Coffee Farms, confirming that the system’s structure naturally sustains cooperation without
sanctions. Bifurcation diagrams identify the critical policy thresholds for reputational benefits), demonstrating that
Vietnam’s reputation incentive program triggers a self-reinforcing cycle: strong local enforcement enables national
policy scaling. Together, these graphs form an irrefutable evidence chain: the dynamic path (time series), structural
stability (phase portrait), and policy leverage points (bifurcation) collectively prove that mutualistic economic
incentives drive scalable sustainability. This trio is not merely illustrative; it is the methodological bedrock of our
contribution to environmental governance theory.

8. Conclusion
This study demonstrates that mutualistic economic incentives, not regulatory coercion, drive stable cooperation in
Vietnamese agro-ecological networks. The evolutionary game-theoretic model, rigorously calibrated to empirical
profit data, reveals a globally stable equilibrium where Fruit Orchards, Bee Apiaries, and Coffee Farms adopt full
cooperation. Government incentives and reputational considerations are critical determinants of the strategic choices
made by both regulators and agricultural producers. High reputational benefits encourage the government to maintain
a strict regulatory stance, promoting full cooperation among all parties and guiding the system toward a Pareto-
optimal outcome. In contrast, when reputational rewards are low or when fiscal commitments to subsidize
cooperation become excessive, the government tends to adopt a lax regulatory posture, even as the other parties may
still pursue cooperative strategies.

Discussion
A critical insight is the hierarchical sensitivity: Subnational Governance is the most responsive to initial conditions,
reflecting its role as a mediator between economic actors and policy. In contrast, producers (Fruit Orchards, Bee
Apiaries, Coffee Farms) exhibit near-universal convergence to full cooperation due to the dominance of mutualistic
payoffs. This hierarchy mirrors real-world dynamics where farmers respond to market incentives, while governance
adapts to stakeholder pressures. The model’s simplicity relying solely on economic parameters highlights the
sufficiency of mutualism in driving sustainability, obviating the need for complex institutional mechanisms. Future
work should explore spatial heterogeneity and climate resilience.

The model's dynamics for Subnational Environmental Governance (v) reveal a nuanced relationship between
reputational benefits (D) and regulatory intensity, which directly challenges conventional governance frameworks
that prioritize top-down enforcement over reputation-based incentives. Our simulation shows that even with a
moderate reputational benefit (𝐷 = 0.5), governance entities converge to a stable equilibrium at 𝑣 ≈ 0.97, not full
enforcement (𝑣 = 1), indicating that reputation alone cannot fully overcome administrative costs and political trade-
offs. This finding aligns with empirical evidence from Vietnamese case studies: while agencies in Dak Lak province
gained public trust (reputation) for enforcing coffee-fruit intercropping regulations, they still faced unavoidable costs
preventing complete regulatory adoption [6]. Crucially, this partial equilibrium is not a failure but a strategic
adaptation to institutional constraints, mirroring real-world governance where agencies balance reputation gains
against operational realities.
Policy Implication
To maximize the delegation effect, countries can learn from Vietnam policies such as reputation incentive through
public recognition programs as demonstrated in Lam Dong province [113]; [114]. This would push 𝑣 toward 1,
enabling the central government to scale policies rather than imposing them. The model thus proves that reputation is
the most cost-effective governance tool, not for enforcing rules, but for enabling scalable, self-sustaining policy
networks.

Another notable policy from Vietnam is the Beekeeping for Pollination Subsidy (under the National Agricultural
Development Program) operationalizes our model's mutualistic parameters by providing supportive subsidies for
beekeeping infrastructure [94]. This directly mirrors our parameterization of pollination benefits and nectar access,
with transformative results.

Application
Building on the proven success of coffee-fruit intercropping systems in Southeast Asia, this model can be
strategically adapted to Central and South America (with Brazil as a flagship case) and Africa/Middle East (with
Ethiopia as a pilot example), leveraging region-specific ecological and agricultural knowledge to enhance
biodiversity, farmer resilience, and pollinator health.

Brazil’s Minas Gerais region offers a robust empirical foundation for scaling this model, demonstrated by the Minas
Gerais trial. This scientifically validated system integrates Coffea arabica with Myrciaria jaboticaba (jaboticaba) and
Psidium guajava (guava) without nutrient competition, a critical advancement over conventional intercropping.
Crucially, spatial root segregation prevents resource conflict: coffee develops deep taproots while jaboticaba and
guava maintain shallow root systems [21]; [43]; [22]. This vertical partitioning eliminates nutrient rivalry, as
confirmed by soil nutrient analyses [9]. The model can be further developed by applying bees into the agroforestry
model. Particularly, the floral phenology of the fruit species creates a year-round nectar/pollen resource for bees:
jaboticaba’s dual flowering peaks (August-September) [8], while guava’s concentrated wet-season bloom (May-July)
bridges the critical gap when coffee flowering wanes [82]. This synergy not only supports bee nutrition but also
enhances coffee pollination efficiency, directly boosting yield stability, a key factor for smallholder adoption in
Brazil’s coffee heartland.

Ethiopia presents a high potential yet underexploited opportunity to implement the SAFE model, building on its
existing agroforestry traditions. While current research lacks a dedicated coffee-fruit intercropping framework
(unlike Brazil’s empirical evidence), Ethiopia’s agroecological context offers fertile ground for adaptation. Existing
studies confirm coffee’s compatibility with Tamarindus indica (tamarind) without nutrient competition [102], and
tamarind’s spring/summer bloom [70] aligns with Ethiopia’s critical dry-season forage gap, ensuring sustained bee
forage when coffee flowering is sparse. This is another research approach beside Ethiopia’s current focus on coffee-
shade tree systems such as Eucalyptus and Inga or coffee-food crop rotations such as Phaseolus vulgaris, which lack
direct pollinator support from bees [23]; [64]. The paper proposes integrating tamarind as a bee-friendly fruit crop,
complementing Ethiopia’s established coffee-korarima model [84] to create a three-tiered system: coffee (canopy),
tamarind (mid-layer), and bees (pollinators). This approach addresses a critical research gap: while Ethiopian studies
on coffee-bee interactions remain limited [2], agroforestry frameworks for coffee-shade integration are mature [1],
providing a scaffold for rapid model deployment. By prioritizing locally adapted species like tamarind, the model
ensures cultural acceptability and scalability, transforming Ethiopia’s coffee landscapes into resilient, pollinator-
supportive ecosystems.

Limitation
While this study advances understanding of sustainable agricultural systems through the SAFE model, several
limitations warrant acknowledgment. These reflect deliberate scope boundaries, data constraints, and methodological
trade-offs inherent in complex sustainability modeling.

Simplified Economic Assumptions and Scope Boundaries

This work intentionally prioritizes environmental and social sustainability dimensions over detailed economic
analysis, consistent with its core focus on agricultural resilience. Consequently, explicit cost structures were not
integrated into the model. This omission is justified by two factors: (i) the primary research objective centered on
sustainability metrics rather than profitability optimization, and (ii) in the Vietnamese context of this study, land
availability and crop input costs are relatively low compared to industrialized agricultural systems, reducing the
immediate sensitivity of sustainability outcomes to fine-grained cost variations. Future work should integrate region-
specific cost modules to enhance economic realism.

Model Abstraction and Parameterization

To ensure computational tractability and focus on core sustainability mechanisms, the model necessarily abstracts
from the full complexity of real-world agricultural systems. Certain variables (e.g., microclimate fluctuations, hyper-
local market shocks, or granular soil heterogeneity) were excluded due to data scarcity and the need for model
parsimony. While this simplification enables robust analysis of primary drivers related to pollination or
environmental symbiosis, it may underrepresent edge-case scenarios. Sensitivity analyses confirmed that these
omissions did not alter the qualitative direction of our key sustainability findings, though precise quantitative
outcomes could vary under extreme, unmodeled conditions.

Operationalization of Intangible Metrics

Critical sustainability dimensions like "reputational value" remain challenging to quantify empirically. In this study,
such parameters were treated as numerical proxies due to the absence of standardized, scalable measurement
frameworks for these constructs in smallholder agricultural contexts. While this approach enabled their inclusion in
the simulation, it introduces uncertainty regarding the precise weighting and behavioral impact of these factors. We
urge future research to develop validated metrics for socio-cultural dimensions in agricultural sustainability to refine
such models.
Reference

1. Abebe T (2005) ‘Diversity in homegarden agroforestry systems of Southern Ethiopia’, ISBN 90-8504-163-5

2. Abrha H (2019) ‘Climate change impact on coffee and the pollinator bee suitable area interaction in Raya
Azebo, Ethiopia’, Cogent Food & Agriculture, 4(1), DOI: 10.1080/23311932.2018.1564538
3. Anh N (2023) ‘Raising bees for honey, raising fruit orchards’, Nong nghiep va Moi truong,
[Link]

4. Anh T (2025) ‘Model of growing fruit trees in coffee gardens’, Thoi Bao Ngan Hang,
[Link]

5. Aparisi A M (2021) ‘Ethiopian coffee marketing reforms and smallholder coffee producers’, iied,
[Link]

6. Bach T (2025) ‘Developing green and ecological agriculture: Resolving policy problems’, Ha Noi moi,
[Link]

7. Bao P (2022) ‘Allowing fruit trees to "coexist peacefully" with coffee trees, farmers suddenly have a large
amount of extra money.’, Dan Viet, [Link]
[Link]

8. CABI (2019) ‘Myrciaria cauliflora (jaboticaba)’, [Link]

9. Campanha MM, Santos RHS, Freitas GB and Finger F (2004) ‘Growth and yield of coffee plants in
agroforestry and monoculture systems in Minas Gerais, Brazil’, Agroforestry Systems, 63(1):75-82, DOI:
10.1023/B:AGFO.0000049435.22512.2d

10. Cesare D M, Perel P, Taylor S, Kabudula C, Bixby H, Gaziano TA, Vaca McGhie D, Mwangi J, Pervan B,
Narula J, Pineiro D, Pinto F J (2024) ‘The Heart of the World’, Glob Heart, 19(1), doi: 10.5334/gh.1288

11. Chemura A, Mudereri BT, Yalew AW and Gornott C (2021) ‘Climate change and specialty coffee potential
in Ethiopia’, Scientific Reports, 11, [Link]

12. Chi K (2021) ‘Violations of regulations on livestock waste treatment are subject to fines of up to 20 million
VND.’, Bao Dien tu Chinh Phu, [Link]
[Link]#:~:text=

13. Dang H (2020) ‘High income from intercropping oranges in coffee gardens’, Lam Dong,
[Link]

14. Dias CG, Martins FB and Martins MA (2024) ‘Climate risks and vulnerabilities of the Arabica coffee in
Brazil under current and future climates considering new CMIP6 models’, Science of The Total Environment,
907, [Link]

15. Fahad S, Chavan S B, Chichaghare A R, Uthappa A R, Kumar M, Kakade V, Pradhan A, Jinger D, Rawale
G, Yadav D K, Kumar V, Farooq T H, Ali B, Sawant A V, Saud S, Chen S and Poczai P (2022) ‘Agroforestry
Systems for Soil Health Improvement and Maintenance’, Sustainability, 14(22),
[Link]

16. FAO (2018) ‘Small family farms country factsheet’, FAO,

[Link]
1b127ff7b458/content#:~:text=Almost%2040%20percent%20of%20the,are%20situated%20in%20rural%20area
s.

17. FAO (2024) ‘Urban and peri-urban agriculture‘, FAO, [Link]

18. FAO (n.d.) ‘Avocado Production in Vietnam - Nguyen Minh Chau & Vo The Truyen*’, FAO,
[Link]

19. FAO (n.d.) ‘Other Fruits, Banana’, FAO, [Link]

20. FAO (n.d.) ‘Potential Fruits’, FAO, [Link]

21. FAO (n.d.) 'Choosing a site and preparing the plantation’, FAO,
[Link]

22. Feedipedia (2016) ‘Guava’, Animal feed resources information system,

[Link]

23. Getahun T, Mamo G, Haile G, Tesfaye G and Landaverde R (2024) ‘Beyond genetics × environment
interaction on quality drivers of Arabica coffee: a review of implication for Gedeo indigenous agroforestry
systems under changing climates’, Beverage Plant Research, 4(1):1-12, DOI: 10.48130/bpr-0024-0032

24. GiaCaPhe (2023) ‘Effectiveness of coffee intercropping model in Dak Lak’, GiaCaPhe,
[Link]

25. Ha N and Dung Q (2025) ‘Central Highlands people continuously "win" coffee season’, Cong An,
[Link]

26. Hai T (2025) ‘Coffee flower season attracts tourists to Dak Lak’, VTV, [Link]
[Link]#:~:text=

27. Hong C (2020) ‘Dak Lak: Economic development thanks to intercropping model’, Lien minh Hop tac xa
Viet Nam, [Link]

28. Hu F B (2025) ‘Diet strategies for promoting healthy aging and longevity: An epidemiological perspective’,
J Intern Med, 295(4):508-531. doi: 10.1111/joim.13728
29. Huong G (2025) ‘Bac Giang: Effective honey beekeeping models’, So Nong nghiep va Moi truong Bac
Giang, [Link]
qua-cac-mo-hinh-nuoi-ong-lay-mat?inheritRedirect=false

30. Huong L (2024) ‘Dak Lak agriculture becomes more stable thanks to multiple pillars’, VOV,
[Link]

31. Huynh Q (2018) ‘Surprising results from the intercropping model of planting bananas alongside durian.’,
Nong Thon moi, [Link]
[Link]

32. IDH (n.d.) ‘Central Highlands, Vietnam’, IDH, [Link]

highlands-vietnam/

33. Iftikhar A, Nausheen R, Muzaffar H, Naeem M A, Farooq M, Khurshid M, Almatroudi A, Alrumaihi F,

Allemailem K S and Anwar H (2022) ‘Potential Therapeutic Benefits of Honey in Neurological Disorders: The
Role of Polyphenols’, Molecules, 27(10):32-97, doi: 10.3390/molecules27103297

34. Khue A (2024) ‘Investing in transport infrastructure in the Mekong Delta to adapt to climate change and
rising sea levels’, VnEconomy, [Link]
[Link]

35. Khuyennong (2025) ‘Techniques for planting and caring for Thai Jackfruit to bear fruit all year round’, So
Nong Nghiep va Moi Truong, [Link]
ra-trai-quanh-nam#:~:text=

36. Lam P (2024) ‘Double profits from intercropping model’, Lam Dong, [Link]
[Link]

37. Lan A (2025) ‘Exploiting advantages from long-term industrial crops’, Dak Lak, [Link]
[Link]/kinh-te/202504/huyen-cu-kuin-khai-thac-loi-the-tu-cay-cong-nghiep-dai-ngay-0cc1d1a/

38. Lan A and Dang T (2024) ‘Zero-waste circular agricultural production helps protect the environment,’
Thanh Hoa, [Link]
[Link]

39. Landaverde R, Rodriguez M T, Parrella J A (2023) ‘Honey Production and Climate Change: Beekeepers’
Perceptions, Farm Adaptation Strategies, and Information Needs’, Insects, 14(6), doi: 10.3390/insects14060493

40. Le H (2025) ‘Coffee flower honey harvesting season in the Central Highlands’, Bien Phong
[Link]
41. Le P N (2025) ‘Billionaire Binh Dinh is a farmer who grows fruit trees intercropped with other trees. There
are some trees that bear fruit that cannot be stopped.’, Dan Viet, [Link]
[Link]

42. Liu W, Dai X and Xu J (2015) ‘Influences of leaf-mining insects on their host plants: A review’,
Collectanea Botanica, DOI: 10.3989/collectbot.2015.v34.005

43. Love K and Paull RE (2011) ‘Jaboticaba’, Fruits and Nuts,

[Link]

44. Mafra D, Ugochukwu SA, Borges NA, Cardozo LFMF, Stenvinkel P and Shiels PG (2024) ‘Food for
healthier aging: power on your plate’, Crit Rev Food Sci Nutr, 64(3):603-616, doi:
10.1080/10408398.2022.2107611.

45. Mai A (2022) ‘Building 20 models of growing fruit trees intercropped with coffee’, Lam Dong,
[Link]

46. Mai V Q (2023) ‘Benefits of three-storey ecological coffee garden’, Cong Thuong,
[Link]

47. Mamuye M, Gallemore C, Jespersen K, Kasongi N, Berecha G (2024) ‘Changing rainfall and temperature
trends and variability at different Spatiotemporal scales threaten coffee production in certain elevations’,
Environmental Challenges, 15, [Link]

48. Marathon Baker (n.d.) ‘Coffee Crunchy Granola Cereal’, [Link]

crunchy-granola-cereal/

49. Mariotti A (2015) ‘The effects of chronic stress on health: new insights into the molecular mechanisms of
brain–body communication’, Future Sci OA, 1(3), doi: 10.4155/fso.15.21

50. Mendoza M F, Sulague R M, Posas-Mendoza T, Lavie C J (2023) ‘Impact of Coffee Consumption on

Cardiovascular Health’, Ochsner J, 23(2):152–158, doi: 10.31486/toj.22.0073

51. Minh Q and Minh H (2023) ‘Sustainability from landscape coffee model’, Nong nghiep va Moi truong,
[Link]

52. Mosavat M, Ooi F K and Mohamed M (2014) ‘Stress Hormone and Reproductive System in Response to
Honey Supplementation Combined with Different Jumping Exercise Intensities in Female Rats’, Biomed Res
Int, doi: 10.1155/2014/123640
53. Munirwan RP, Taib AM, Taha MR, Rahman NA and Munirwansyah M (2022) ‘Utilization of coffee husk
ash for soil stabilization: A systematic review’, Physics and Chemistry of the Earth, Parts A/B/C, 128,
[Link]

54. Muwal N (2022) ‘Life Cycle Of The Coffee Bean’, Long Bottom Coffee,
[Link]
bean#:~:text=There%20is%20usually%20one%20major,last%20four%20to%20six%20months.

55. Nam H (2025) ‘Prospects of fruit tree intercropping model in coffee gardens’, Dak Lak Dien Tu,
[Link] [Link]/kinh-te/202505/trien-vong-mo-hinh-trong-xen-cay-an-trai-trong-vuon-ca-phe-84e1ca7/

56. Nguyen D (2014) ‘Beekeeping in the middle of a coffee garden’, Khoa Hoc
[Link]

57. Nguyen G (2016) ‘Get rich from the model of intercropping fruit trees in coffee gardens’, Dak Lak,
[Link] [Link]/channel/3483/201607/lam-giau-tu-mo-hinh-trong-xen-cay-an-trai-trong-vuon-ca-phe-
2442885/

58. Nguyen L (2016) ‘Intercropping model achieves high economic efficiency’, Lam Dong,
[Link]

59. Nguyen P (2020) ‘Intercropping in coffee gardens brings high efficiency’, Nhan Dan,
[Link]

60. Nguyen T Y (2025) ‘The results of the 3-year implementation of the Project: "Developing and promoting
intercropping models for a number of fruit trees in coffee gardens in the Central Highlands" in Lam Đong
Province.’, So Nong Nghiep va Moi Truong tinh Lam Dong, [Link]
khuyen-nong/chuyen-giao-khkt/3248-

61. Nguyen TQ, Tran TDH, Duong TO, Nguyen QN, Nguyen TN and Landaverde R (2019) ‘Flower biology of
black pepper (Piper nigrum) in Vietnam, Journal of Vietnam Agricultural Science and Technology, 1(4),
[Link]

62. Nguyen TTT, Khoi DN, Denis A, Viet LV, Wellens J and Tychon B (2022) ‘Early Prediction of Coffee
Yield in the Central Highlands of Vietnam Using a Statistical Approach and Satellite Remote Sensing
Vegetation Biophysical Variables’, Remote Sens, 14(13), [Link]

63. Nguyet A (2025) ‘Income over 100 million VND/year from beekeeping model for honey and fruit tree
planting’, Cao Bang, [Link]
64. Nigussie W, Al-Najjar H, Zhang W, Yirsaw E, Nega W, Zhang Z and Kalantar B (2024) ‘Enhancing Coffee
Agroforestry Systems Suitability Using Geospatial Analysis and Sentinel Satellite Data in Gedeo Zone,
Ethiopia’, Sensors, 24(19), [Link]

65. NLSH (2023) ‘What month does avocado tree bloom? Off-season flowering techniques’, NLSH,
[Link]
may/?srsltid=AfmBOorQnwsDxfxqHhWO05qPjj8SrHQzlTNi4_bd7IkqVFYhuYw58q0Y

66. OEC (n.d.) ‘Coffee in Ethiopia’, OEC[dataset], [Link]

product/coffee/reporter/eth

67. OEC (n.d.) ‘Bananas & Plantains (Fresh/Dried) in Vietnam’, OEC[dataset],

[Link]

68. OEC (n.d.) ‘Coffee in Vietnam’, OEC[dataset], [Link]

product/coffee/reporter/vnm

69. OEC (n.d.) ‘Pepper in Vietnam’, OEC[dataset], [Link]

product/pepper/reporter/vnm

70. Orwa (2009) ‘Tamarindus indica’, Agroforestry Database 4.0,

[Link]

71. Pandian K, Mustaffa MR, Mahalingam G, Paramasivam A, Prince AJ, Gajendiren M, Mohammad ARR and
Varanasi ST (2024) ‘Synergistic conservation approaches for nurturing soil, food security and human health
towards sustainable development goals’, Journal of Hazardous Materials Advances, 16,
[Link]
0809

72. Phuc T and Hoang L (2021) ‘Model of intercropping fruit trees in coffee gardens in Dak Ha’, Nhan Dan,
[Link]

73. Phuong L (2024) ‘Double profits from intercropping model’, Lam Dong [Link]
[Link]

74. Phuong T and Hang B (n.d.) ‘The effectiveness of the intercropping model of growing avocado in coffee
gardens.’ So Nong nghiep va Moi truong Lam Dong, [Link]
gioi/1199-hieu-qua-tu-mo-hinh-trong-xen-bo-trong-vuon-ca-phe
75. Poncet V, van Asten P, Millet CP, Vaast P, Allinne C and Landaverde R (2024) ‘Which diversification
trajectories make coffee farming more sustainable?, Current Opinion in Environmental Sustainability, 68,
[Link]

76. Poovizhiraja B, Ambiga J, Divya C, Malathi J, Poonguzhali S, Mekala P (2023) ‘Role of Honey Bees in
Horticultural Crops’, Just Agriculture, 3(12), e-ISSN: 2582-8223

77. Pukka (n.d.) ‘Lemon, Ginger & a touch of Manuka Honey’, Pukka,
[Link]

78. Quoc H (2022) ‘The effectiveness of the fruit tree growing model combined with beekeeping’, Van hoa va
Doi song, [Link]

79. Rahman A, Baharlouei P, Koh E H Y, Pirvu D G, Rehmani R, Arcos M and Puri S (2024) ‘A
Comprehensive Analysis of Organic Food: Evaluating Nutritional Value and Impact on Human Health’, Foods,
13(2), [Link]

80. Ramamoorthy K, Dhanraj R, Vijayakumar N, Ma Y, Obaid A S and Narayanan M (2024) ‘Vegetable and
fruit wastes: Valuable source for organic fertilizer for effective growth of short-term crops: Solanum
lycopersicum and Capsicum annum’, Environmental Research, 251,
[Link]

81. Samnegård U (2020) ‘Why is arabica coffee visited by so few non-Apis bees in its native range?”, Ecology,
9:(101), [Link]

82. Sandhu MS, Kaur R and Bal JS (2024) ‘Studies on floral bud development and flowering behaviour of
guava (Psidium guajava L.)’, International Journal of Pharmacy and Pharmaceutical Science, 6(1):38-43, DOI:
[Link]

83. SenGupta S and Atal A (2024) ‘Income inequality in the face of climate change: an empirical investigation
on unequal nations, vulnerable regions and India’, SN Business & Economics, 4(87),
[Link]

84. Seyum EG, Mekonnen B, Gebre GH and Kebebew ST (2020) ‘Intercropping Coffee (Coffea arabica) and
Korarima (Aframomum corrorima (Braun) P.C.M. Jansen) at Tepi, Southwest Ethiopia’, American Journal of
Agriculture and Forestry, 8(4):175-180, DOI: 10.11648/[Link].20200804.20

85. Simexco Dak Lak (2024) ‘Sharing the secret of intercropping pepper with coffee to help increase
productivity’, Simexco Dak Lak, [Link]
86. Thanh N (2023) ‘Get rich from growing fruit trees and raising bees for honey’, Dan toc va Phat trien,
[Link]

87. Thanh T and Thanh T (2025) ‘Honey beekeeping model for high economic efficiency’, Thanh Hoa,
[Link]

88. Thu N (2018) ‘The effectiveness of intercropping method in Ea Rieng commune, M'Drak district’, Cong
Thong tin Dien tu tinh Dak Lak, [Link] [Link]/tin-tuc-cac-so-ban-nganh/-
/asset_publisher/bDngMUmMrWIw/content/hieu-qua-phuong-phap-xen-canh-cay-trong-o-xa-ea-rieng-huyen-
m-rak

89. Tin (n.d.) ‘Intercropping measures must comply with technical processes to develop sustainable Robusta
coffee.’,Hoi Nong dan, [Link] [Link]/moi-truong/-
/asset_publisher/bDngMUmMrWIw/content/bien-phap-xen-canh-cac-loai-cay-trong-an-tuan-thu-qui-trinh-ky-
thuat-e-phat-trien-ca-phe-voi-ben-vung/pop_up?_101_INSTANCE_bDngMUmMrWIw_viewMode=print

90. Tran V H (2025) ‘Durian production area in association with seasons in Vietnam’, Nong nghiep moi truong,
[Link]
[Link]

91. Trung H (2024) ‘"Bringing forests to coffee gardens" - opening a sustainable direction for the Vietnamese
coffee industry’, Nhan Dan [Link]
[Link]

92. Tuan H (2025) ‘Get rich from growing coffee combined with fruit trees’, Bao Son La,
[Link]

93. TVPL (2023) ‘On the approval of the project "Intercropping and livestock farming under cashew tree
canopies, phase 2023–2025"’, Thu Vien Phap Luat, [Link]
[Link]

94. TVPL (2024) ‘Approval of the proposal "Sustainable Development of the Beekeeping Industry towards
2030" by the Minister of Agriculture and Rural Development’, Thu Vien Phap Luat,
[Link]
[Link]

95. Valentina (n.d.) ‘Valentina Coffee Flower Pollens’, [Link]

flower-pollens-320g/
96. Van L (2021) ‘Lam Dong: Raising safe biological bees in a coffee garden, selling both milk and honey,
many people come to see’, Gia Lai, [Link]
[Link]

97. Van V (n.d.) ‘Stories from Lam Đong: Thousands of beehives "on the same path."’, So Nong Nghiep va Moi
Truong tinh Lam Dong, [Link]

98. VietGap (n.d.) ‘VietGap’, [Link]

99. Vigami (2024) ‘What month does durian tree bloom?’, Vigami, [Link]
ra-hoa-vao-thang-may/#:~:text=

100. VTV (2018) ‘400 hectares of coffee for 20 years without pesticides’, VTV, [Link]
[Link]

[Link] K, Cakmak I, Wang S, Zhang F and Guo S (2021) ‘Synergistic and antagonistic interactions between
potassium and magnesium in higher plants’, The Crop Journal, 9(2):249-256,
[Link]

102. Zeleke G, Dejene T, Tadesse W and Martín-Pinto P (2021) ‘Land-Use Impact on Stand Structure and Fruit
Yield of Tamarindus indica L. in the Drylands of Southeastern Ethiopia’, Life (Basel), 11(5), doi:
10.3390/life11050408

103. United Nations (n.d.) ‘The 17 Goals’, United Nations, [Link]

104. Yu L, Hu X, Zhang Q and Zheng M (2017) ‘Improving Agricultural Water Use Efficiency: A Quantitative
Study of Zhangye City Using the Static CGE Model with a CES Water−Land Resources Account’,
Sustainability, 9(2), DOI: 10.3390/su9020308

105. Li Y, Sun Z, Accatino F, Hang S, Lv Y and Ouyang Z (2021) ‘Comparing specialised crop and integrated
crop-livestock systems in China with a multi-criteria approach using the emergy method’, Journal of Cleaner
Production, 314, [Link]

106. Huo H and Liu X (2024) ‘Behavioral decision-making of government, agricultural product producers, and
consumers on agricultural product quality and safety regulation in a digital environment’, Front. Public Health,
12, [Link]

107. Xi X and Zhang J (2020) ‘Complexity analysis of a decision-making game concerning governments and
heterogeneous agricultural enterprises with bounded rationality’, Chaos, Solitons & Fractals, 140,
[Link]
108. Gong M, Zhong Y, Elahi E, Han F and Chen Y (2025) ‘Evolutionary game mechanisms for food savings
and loss reduction in the grain industry’, Journal of the Science of Food and Agriculture,
[Link]

109. Zhang W and Liu H (2024) ‘Evolutionary game analysis of multi-agent cooperation strategy analysis in
agricultural water conservancy PPP project under digitization background’, Scientific Reports, 14,
[Link]

110. Shen X, Tang J, Zhang Y, Qian B, Li J, Zhou M, Zhao Y and Yin Y (2024), ‘Dynamic Evolution Game
Strategy of Government, Power Grid, and Users in Electricity Market Demand-Side Management’,
Mathematics, 12(20), [Link]

111. Chen Y, Sun Z, Zhou Y, Tang W and Ma Y (2024) ‘The future of sustainable farming: An evolutionary
game framework for the promotion of agricultural green production technologies’, Journal of Cleaner
Production, 460, [Link]

112. Wang F and Xu Y (2022) ‘Evolutionary Game Analysis of the Quality of Agricultural Products in Supply
Chain’, Agriculture, 12(10), [Link]

113. Thao H (2023) ‘Awarding ceremony of Green Startup Project Competition’, Nhan Dan,
[Link]

114. Bao Chinh Phu (2021) ‘National Green Growth Strategy for 2021-2030, vision towards 2050’, Government
News, [Link]
[Link]#:~:text=

115. Hieu M (2024) ‘Vietnam durian exports to China surpass Thailand’s’, VnExpress,
[Link]
[Link]

116. Vu H (2019) ‘Opportunity to export Vietnamese avocado to the US’, Cong Thuong,
[Link]

117. Nguyen H (2024) ‘Opportunity to boost avocado and passion fruit exports to the Chinese market’,
VnBusiness, [Link]
[Link]?fbclid=IwAR2gDKlNVQzP1wlaboXHdKsQaow2WcuQgx7Ed-
fOqQv6rBm3ljOPKVz4fKc

118. Ha H (2022) ‘Vietnam exports 3 tons of frozen avocados to ‘avocado-producing powerhouse’ Australia’,
The Gioi va Viet Nam, [Link]
[Link]
 119. Khuat DL, Le XH, Hoa TD and Phong HP (2012) ‘A preliminary study on bees (Hemenoptera: Apoidea:
Apiformes) from Northern and North central Vietnam’, TAP CHI SINH HOC, 34(4):414-421, DOI:
10.15625/0866-7160/v34n4.2676

120. Minh Q and Dang L (2025) ‘Durian growing area in Dak Lak will be the largest in the country’, Nong
nghiep va Moi truong, [Link]
[Link]

121. FSIA (n.d.) ‘Analysis of the Dak Lak Avocado Chain’, Fresh Studio Innovations Asia,
[Link]

122. Duong T O, Nguyen V L, Nguyen Q N, Tran T D H, Pham T H and Nguyen B H (2021) ‘Assessment of
Growth and Productivity on Four Black Pepper Varieties (Piper nigrum L.) in Three Target Provinces of
Vietnam’, Engineering, 13(12), DOI: 10.4236/eng.2021.1312046

123. Tran DM and Nguyen T H (2024) ‘Endophytic bacterial dataset of the Cavendish banana grown in Dak Lak
Province of Vietnam using 16S rRNA gene metabarcoding’, Data in Brief, 52,
[Link]

124. Friedman D. (1991) ‘Evolutionary Games in Economics.’ Econometrica, 59(3):637-666.

[Link]

Appendix

% Parameters
K_C_A = 0.3; K_A_B = 0.4; K_A_C = 0.5;
L_C_A = 0.2; L_C_B = 0.3; L_A_C = 0.4;
O_C_A = 0.5; O_A_C = 0.3;
M_B_A = 0.4; M_B_C = 0.5;
G_A = 0.4; G_B = 0.4; G_C = 0.4;
H_A = 0.6; H_B = 0.6; H_C = 0.6;
I_A = 0.3; I_B = 0.3; I_C = 0.3;
J_A = 0.3; J_B = 0.3; J_C = 0.3;
E_A = 0.2; E_B = 0.2; E_C = 0.2;
D = 0.5; D_A = 0.5; D_B = 0.5; D_C = 0.5;
% Time span
tspan = [0 1]; % Extended time span
% Different initial conditions
initial_conditions = { ...
[0.9; 0.9; 0.9; 0.9], ... % High cooperation
[0.1; 0.9; 0.9; 0.9], ... % Low governance, high cooperation
[0.5; 0.5; 0.5; 0.5], ... % Balanced
[0.2; 0.2; 0.2; 0.2], ... % Low cooperation
[0.7; 0.3; 0.3; 0.3] ... % High governance, low cooperation
};
% Colors
colors = {'b', 'r', 'g', 'm', 'c'};
% Dynamics
dynamics = @(t, u) [
u(1) * (1 - u(1)) * (u(2) * (-G_A - H_A - I_A - J_A - E_A) + u(3) * (-
G_B - H_B - I_B - J_B - E_B) + u(4) * (-G_C - H_C - I_C - J_C - E_C) + (2*D
+ I_A + J_A + I_B + J_B + I_C + J_C + E_A + E_B + E_C ));
u(2) * (1 - u(2)) * (u(1) * (G_A + H_A + I_A + J_A) + u(3) * (2*M_B_A )
+ u(4) * (2*K_C_A + 2*L_C_A + 2*O_C_A) + 2*D_A);
u(3) * (1 - u(3)) * (u(1) * (G_B + H_B + I_B + J_B) + u(2) * (2*K_A_B) +
u(4) * (2*L_C_B) + 2*D_B);
u(4) * (1 - u(4)) * (u(1) * (G_C + H_C + I_C + J_C) + u(2) * (2*K_A_C +
2*L_A_C + 2*O_A_C) + u(3) * (2*M_B_C) + 2*D_C)
];
% Time series plots
figure;
subplot(2, 2, 1);
hold on; grid on;
xlabel('Time (t)'); ylabel('Probability (v)'); title('Subnational
Governance');
legend_str = {};
for i = 1:length(initial_conditions)
ic = initial_conditions{i};
[t, sol] = ode45(dynamics, tspan, ic);
v_t = sol(:, 1);
 plot(t, v_t, 'LineWidth', 2, 'Color', colors{i});
legend_str{end+1} = sprintf('IC_%d: (%.1f,%.1f,%.1f,%.1f)', i, ic(1),
ic(2), ic(3), ic(4));
end
legend(legend_str, 'FontSize', 8);
subplot(2, 2, 2);
hold on; grid on;
xlabel('Time (t)'); ylabel('Probability (x)'); title('Fruit Orchards');
for i = 1:length(initial_conditions)
ic = initial_conditions{i};
[t, sol] = ode45(dynamics, tspan, ic);
x_t = sol(:, 2);
plot(t, x_t, 'LineWidth', 2, 'Color', colors{i});
end
subplot(2, 2, 3);
hold on; grid on;
xlabel('Time (t)'); ylabel('Probability (y)'); title('Bee Apiaries');
for i = 1:length(initial_conditions)
ic = initial_conditions{i};
[t, sol] = ode45(dynamics, tspan, ic);
y_t = sol(:, 3);
plot(t, y_t, 'LineWidth', 2, 'Color', colors{i});
end
subplot(2, 2, 4);
hold on; grid on;
xlabel('Time (t)'); ylabel('Probability (z)'); title('Coffee Farms');
for i = 1:length(initial_conditions)
ic = initial_conditions{i};
[t, sol] = ode45(dynamics, tspan, ic);
z_t = sol(:, 4);
plot(t, z_t, 'LineWidth', 2, 'Color', colors{i});
end
sgtitle('Time Series: Convergence with Percentage-Based Parameters and
Extended Time Span');
>>

Code for time plot

% Time span
tspan = [0 100];
% Multiple random initial conditions to show basins of attraction
n_trajectories = 50;
rng(1); % For reproducibility
initial_conditions = rand(n_trajectories, 4);
% Dynamics
dynamics = @(t, u) [
u(1) * (1 - u(1)) * (u(2) * (-G_A - H_A - I_A - J_A - E_A) + u(3) * (-
G_B - H_B - I_B - J_B - E_B) + u(4) * (-G_C - H_C - I_C - J_C - E_C) + (2*D
+ I_A + J_A + I_B + J_B + I_C + J_C + E_A + E_B + E_C ));
u(2) * (1 - u(2)) * (u(1) * (G_A + H_A + I_A + J_A) + u(3) * (2*M_B_A )
+ u(4) * (2*K_C_A + 2*L_C_A + 2*O_C_A) + 2*D_A);
u(3) * (1 - u(3)) * (u(1) * (G_B + H_B + I_B + J_B) + u(2) * (2*K_A_B) +
u(4) * (2*L_C_B) + 2*D_B);
u(4) * (1 - u(4)) * (u(1) * (G_C + H_C + I_C + J_C) + u(2) * (2*K_A_C +
2*L_A_C + 2*O_A_C) + u(3) * (2*M_B_C) + 2*D_C)
];
% Phase portrait: Fruit Orchards vs Coffee Farms
figure;
hold on;
grid on;
xlabel('Probability of Resource Sharing (x) - Fruit Orchards');
ylabel('Probability of Resource Sharing (z) - Coffee Farms');
title('Phase Portrait: Fruit Orchards vs Coffee Farms');
axis([0 1 0 1]);
% Color map for different outcomes
colormap('jet');
colors = jet(n_trajectories);
for i = 1:n_trajectories
ic = initial_conditions(i, :);
[t, sol] = ode45(dynamics, tspan, ic);
% Extract x and z
x_t = sol(:, 2);
z_t = sol(:, 4);
% Plot trajectory with color based on final state
plot(x_t, z_t, 'LineWidth', 1.5, 'Color', colors(i, :));
 % Mark starting point
plot(x_t(1), z_t(1), 'o', 'MarkerSize', 6, 'Color', colors(i, :));
% Mark endpoint
plot(x_t(end), z_t(end), 's', 'MarkerSize', 6, 'Color', colors(i, :));
end
legend('Trajectories', 'Start points', 'End points');
hold off;

Code for Phase Portrait Coffee-Fruit

% Parameters
K_C_A = 0.3; K_A_B = 0.4; K_A_C = 0.5;
L_C_A = 0.2; L_C_B = 0.3; L_A_C = 0.4;
O_C_A = 0.5; O_A_C = 0.3;
M_B_A = 0.4; M_B_C = 0.5;
G_A = 0.4; G_B = 0.4; G_C = 0.4;
H_A = 0.6; H_B = 0.6; H_C = 0.6;
I_A = 0.3; I_B = 0.3; I_C = 0.3;
J_A = 0.3; J_B = 0.3; J_C = 0.3;
E_A = 0.2; E_B = 0.2; E_C = 0.2;
D_A = 0.5; D_B = 0.5; D_C = 0.5;
% Time span
tspan = [0 2];
% Multiple random initial conditions
n_trajectories = 50;
rng(1);
initial_conditions = rand(n_trajectories, 4);
% Dynamics
dynamics = @(t, u) [
u(1) * (1 - u(1)) * (u(2) * (-G_A - H_A - I_A - J_A - E_A) + u(3) * (-
G_B - H_B - I_B - J_B - E_B) + u(4) * (-G_C - H_C - I_C - J_C - E_C) + (2*D
+ I_A + J_A + I_B + J_B + I_C + J_C + E_A + E_B + E_C ));
u(2) * (1 - u(2)) * (u(1) * (G_A + H_A + I_A + J_A) + u(3) * (2*M_B_A )
+ u(4) * (2*K_C_A + 2*L_C_A + 2*O_C_A) + 2*D_A);
u(3) * (1 - u(3)) * (u(1) * (G_B + H_B + I_B + J_B) + u(2) * (2*K_A_B)
+ u(4) * (2*L_C_B) + 2*D_B);
u(4) * (1 - u(4)) * (u(1) * (G_C + H_C + I_C + J_C) + u(2) * (2*K_A_C +
2*L_A_C + 2*O_A_C) + u(3) * (2*M_B_C) + 2*D_C)
];
% Phase portrait: Subnational Governance vs Fruit Orchards
figure;
hold on;
grid on;
xlabel('Probability of Environmental Regulation (v) - Governance');
ylabel('Probability of Resource Sharing (x) - Fruit Orchards');
title('Phase Portrait: Governance vs Fruit Orchards');
axis([0 1 0 1]);

for i = 1:n_trajectories
ic = initial_conditions(i, :);
[t, sol] = ode45(dynamics, tspan, ic);

% Extract v and x
v_t = sol(:, 1);
x_t = sol(:, 2);

% Plot trajectory
plot(v_t, x_t, 'LineWidth', 1.5, 'Color', colors(i, :));
end
hold off;

% Phase portrait: Bee Apiaries vs Coffee Farms

figure;
hold on;
grid on;
xlabel('Probability of Resource Sharing (y) - Bee Apiaries');
ylabel('Probability of Resource Sharing (z) - Coffee Farms');
title('Phase Portrait: Bee Apiaries vs Coffee Farms');
axis([0 1 0 1]);

for i = 1:n_trajectories
ic = initial_conditions(i, :);
[t, sol] = ode45(dynamics, tspan, ic);
 % Extract y and z
y_t = sol(:, 3);
z_t = sol(:, 4);

% Plot trajectory
plot(y_t, z_t, 'LineWidth', 1.5, 'Color', colors(i, :));
end
hold off;

Code for Phase Portrait Coffee-Bee

% Parameters
K_C_A = 0.3; K_A_B = 0.4; K_A_C = 0.5;
L_C_A = 0.2; L_C_B = 0.3; L_A_C = 0.4;
O_C_A = 0.5; O_A_C = 0.3;
M_B_A = 0.4; M_B_C = 0.5;
G_A = 0.4; G_B = 0.4; G_C = 0.4;
H_A = 0.6; H_B = 0.6; H_C = 0.6;
I_A = 0.3; I_B = 0.3; I_C = 0.3;
J_A = 0.3; J_B = 0.3; J_C = 0.3;
E_A = 0.2; E_B = 0.2; E_C = 0.2;
D_A = 0.5; D_B = 0.5; D_C = 0.5;
% Time span
tspan = [0 2];
% Multiple random initial conditions
n_trajectories = 50;
rng(1);
initial_conditions = rand(n_trajectories, 4);
% Dynamics
dynamics = @(t, u) [
u(1) * (1 - u(1)) * (u(2) * (-G_A - H_A - I_A - J_A - E_A) + u(3) * (-
G_B - H_B - I_B - J_B - E_B) + u(4) * (-G_C - H_C - I_C - J_C - E_C) + (2*D
+ I_A + J_A + I_B + J_B + I_C + J_C + E_A + E_B + E_C ));
u(2) * (1 - u(2)) * (u(1) * (G_A + H_A + I_A + J_A) + u(3) * (2*M_B_A )
+ u(4) * (2*K_C_A + 2*L_C_A + 2*O_C_A) + 2*D_A);
u(3) * (1 - u(3)) * (u(1) * (G_B + H_B + I_B + J_B) + u(2) * (2*K_A_B)
+ u(4) * (2*L_C_B) + 2*D_B);
 u(4) * (1 - u(4)) * (u(1) * (G_C + H_C + I_C + J_C) + u(2) * (2*K_A_C +
2*L_A_C + 2*O_A_C) + u(3) * (2*M_B_C) + 2*D_C)
];
% Phase portrait: Bee Apiaries vs Fruit Orchards
figure;
hold on;
grid on;
xlabel('Probability of Resource Sharing (y) - Bee Apiaries');
ylabel('Probability of Resource Sharing (x) - Fruit Orchards');
title('Phase Portrait: Bee Apiaries vs Fruit Orchards');
axis([0 1 0 1]);
% Color map
colormap('jet');
colors = jet(n_trajectories);
for i = 1:n_trajectories
ic = initial_conditions(i, :);
[t, sol] = ode45(dynamics, tspan, ic);
% Extract y and x
y_t = sol(:, 3);
x_t = sol(:, 2);
% Plot trajectory
plot(y_t, x_t, 'LineWidth', 1.5, 'Color', colors(i, :));
% Mark starting point
plot(y_t(1), x_t(1), 'o', 'MarkerSize', 6, 'Color', colors(i, :));
% Mark endpoint
plot(y_t(end), x_t(end), 's', 'MarkerSize', 6, 'Color', colors(i, :));
end
legend('Trajectories', 'Start points', 'End points');
hold off;
>>

Code for Phase Portrait Fruit-Bee

>> % Simulation time

tspan = [0 2];
% Initial values
y0 = [0.5, 0.5, 0.5, 0.5]; % [Subnational governance, Fruit orchards, Bee
apiaries, and Coffee farms]
% Parameters
K_C_A = 0.3; K_A_B = 0.4; K_A_C = 0.5;
L_C_A = 0.2; L_C_B = 0.3; L_A_C = 0.4;
O_C_A = 0.5; O_A_C = 0.3;
M_B_A = 0.4; M_B_C = 0.5;
G_A = 0.4; G_B = 0.4; G_C = 0.4;
H_A = 0.6; H_B = 0.6; H_C = 0.6;
I_A = 0.3; I_B = 0.3; I_C = 0.3;
J_A = 0.3; J_B = 0.3; J_C = 0.3;
E_A = 0.2; E_B = 0.2; E_C = 0.2;
D_A = 0.5; D_B = 0.5; D_C = 0.5;
% D values to test
D_values = [0.5, 1, 1.5, 2, 2.5];
% Create figure with specific size
figure('Position', [100 100 800 600]);
% Color scheme for different D values
colors = {'b-', 'r-', 'y-', 'm-', 'g-'};
% Loop through different D values
for i = 1:length(D_values)
D = D_values(i);
% Dynamic equation system
dynamics = @(t, u) [
u(1) * (1 - u(1)) * (u(2) * (-G_A - H_A - I_A - J_A - E_A) + u(3) * (-
G_B - H_B - I_B - J_B - E_B) + u(4) * (-G_C - H_C - I_C - J_C - E_C) + (2*D
+ I_A + J_A + I_B + J_B + I_C + J_C + E_A + E_B + E_C ));
u(2) * (1 - u(2)) * (u(1) * (G_A + H_A + I_A + J_A) + u(3) * (2*M_B_A
) + u(4) * (2*K_C_A + 2*L_C_A + 2*O_C_A) + 2*D_A);
u(3) * (1 - u(3)) * (u(1) * (G_B + H_B + I_B + J_B) + u(2) * (2*K_A_B)
+ u(4) * (2*L_C_B) + 2*D_B);
u(4) * (1 - u(4)) * (u(1) * (G_C + H_C + I_C + J_C) + u(2) * (2*K_A_C
+ 2*L_A_C + 2*O_A_C) + u(3) * (2*M_B_C) + 2*D_C)
];
% Solve the system of differential equations
[t, y] = ode45(dynamics, tspan, y0);
% Plot only the v (Subnational governance) probability
plot(t, y(:,1), colors{i}, 'LineWidth', 1.5);
 hold on;
end
% Chart formatting
xlabel('Time (t)');
ylabel('Probability (v)');
legend('D = 0.5', 'D = 1', 'D = 1.5', 'D = 2', 'D = 2.5', 'Location',
'best');
grid on;
axis([0 2 0 1]);
% Adjust appearance
set(gca, 'Box', 'on');
set(gcf, 'Color', 'white');

Code for Bifurcation Analysis for D (Reputation)