Land Use Practices: Sustainability Impacts on Smallholder Farmers

Abstract

This study investigates the drivers of individual and joint adoption of sustainable land use (SLU) practices—specifically crop choice and soil and water conservation—and their impact on farm performance (crop revenue) and production risk (crop yield skewness). Using a farm-level dataset of 504 households across three agro-ecological zones in Punjab, Pakistan, we address selectivity bias through the newly developed multinomial endogenous switching regression (MESR) model. Additionally, we assess land use sustainability across ecological, social, and economic dimensions using a comprehensive non-parametric approach. Our findings identify key determinants of SLU adoption, including farmer education, access to advisory services, FBO membership, hired labor, climate information, farm size, and perceptions of drought and heatwaves. We demonstrate that joint adoption of SLU practices maximizes crop revenue and reduces production risk, lowering the likelihood of crop failure. The study further suggests complementarity between these SLU practices in enhancing crop revenue. Moreover, joint adopters of SLU practices significantly outperform non-adopters in ecological, social, and economic sustainability dimensions. We recommend improving access to public sector farm advisory services and climate information to enable farmers to make well-informed decisions based on reliable data. Implementing these measures can support the transition toward sustainable land management, helping to mitigate risks like crop failure and declining revenues, which threaten farm income.

1. Introduction

Climate change has emerged as one of the most significant challenges to global food security due to unprecedented variability in temperature and rainfall patterns [1]. While Pakistan contributes less than 1% of global GHG emissions [2], it remains one of the most affected countries, as a significant portion of its population directly or indirectly depends on the agriculture sector, which is highly vulnerable to climate change [3]. The agricultural sector’s vulnerability is primarily attributed to institutional, structural, and technological weaknesses, food insecurity, and rural poverty. Furthermore, variations in rainfall patterns, along with frequent droughts and floods, have exacerbated crop failures, productivity losses, and the decline in staple food grains such as wheat and rice in the country [4]. Numerous studies have documented the adverse impacts of climate change on crop productivity in developing countries [5,6]. A core issue is how climate change will negatively impact crop productivity and the potential losses resulting from climate variability and uncertainty. Therefore, adopting sustainable land use (SLU) practices enhances the effectiveness of adaptation and mitigation measures. SLU practices refer to approaches and methods aimed at managing land resources in a way that meets current needs without compromising the ability of future generations to meet their own needs. Although international bodies have strongly recommended the inclusion of appropriate adaptation strategies in national development plans [7,8], the uptake of these practices in Pakistan remains low, posing significant threats to national food security and exacerbating rural poverty.

In developing countries like Pakistan, a deeper understanding of the determinants of SLU practices is crucial for adapting and implementing effective measures to minimize the adverse impacts of climate change on the agricultural sector [9]. Climate change has profoundly impacted Pakistan, exacerbating the country’s socio-economic and geographical vulnerabilities. Over the past two decades, the frequency of natural disasters, including heatwaves, floods, and droughts, has tripled as a result of climate change [4,10]. According to [11], economists increasingly advocate for adaptation strategies as the primary approach to climate change mitigation. Among the most widely documented SLU practices are crop choice—such as changing crops, adopting new varieties, and selecting drought and heat-resistant crop varieties—and soil and water conservation measures, including raised-bed planting, zero tillage, dry sowing of rice, laser land leveling, and the use of organic manure. These practices have been shown to help mitigate the adverse effects of climate change [12,13]. The FAO [14] defines a practice as part of sustainable practices if it (a) has a significant sustainable impact on crop productivity and farm income, (b) helps build resilience to climate change, and (c) reduces GHG emissions. Both crop choice adoption and soil and water conservation practices meet these criteria, making them effective mitigation strategies that may enhance farm income, strengthen the resilience of rural communities, and contribute to lowering GHG emissions.

Numerous studies have examined the individual impact of SLU practices and reported varied findings [12,15,16,17,18,19]. Many of these studies have documented that SLU practices increase crop yields, reduce crop failure, and enhance food security [19,20]. However, other studies have reported that adopting land use practices can reduce crop productivity and farm returns. For instance, [12] found that adopting crop choice and soil and water conservation practices led to higher maize yields. Additionally, studies have highlighted the positive impact of crop choice and water conservation on technical efficiency [21,22]. While existing literature identifies the key determinants of SLU practices and their effects on farm productivity and risk exposure, a research gap remains regarding the potential substitutability or complementarity of adopting individual and joint practices. Moreover, the inconclusive findings concerning the role of SLU practices in crop revenue and downside production risk exposure further highlight the need for deeper investigation. Notably, existing studies overlook the synergy between SLU practices and their subsequent impacts, which is a key contribution of this study to the literature. Additionally, existing studies often overlook the role of agroecology when assessing the effects of SLU practices, a factor addressed explicitly in this study. Moreover, no study has simultaneously examined the determinants of SLU practices and their impacts on ecological, social, and economic dimensions of sustainability.

Several studies have examined the impact of multiple SLU practices on crop productivity and downside risk exposure, often focusing on individual crops such as maize, rice, or wheat [12,23,24]. However, these studies typically approach the issue from a single-crop perspective, which may lead to over-estimations or under-estimations of the outcomes for several reasons. First, SLU practices can benefit subsequent crops in a mixed cropping system, such as wheat or maize, which cannot be fully captured if only rice yield is considered while ignoring subsequent crops. Second, in a mixed cropping system, adopting SLU practices may have adverse effects on subsequent crops, particularly nitrogen-intensive crops like maize and rice, where the yield of one crop may increase at the expense of the other. Therefore, analyzing productivity (crop revenue) over one year is more robust, typically considering three crops, rather than from a monocropping perspective.

This study estimates the joint impact of SLU practices and examines how these practices influence crop revenue and production risk exposure in Pakistan. Notably, we redefine SLU practices to include both crop choice and soil and water conservation practices [1]. The joint adoption of these practices addresses climate change variability, particularly recent events such as heatwaves, droughts, and rainfall fluctuations. Previous studies have explored the link between adopting new crops or altering planting dates as part of climate change mitigation strategies [12,25]. Our study contributes to the existing empirical literature by incorporating the latest advancements in impact assessment methods [26,27], notably by using the MESR model, which better accounts for selection bias in a multinomial framework. This approach allows us to estimate location- and zone-specific factors that impact the adoption of SLU practices and, consequently, the effect of SLU practices on farm performance (crop revenue) and downside production risk exposure among farmers. To our knowledge, this is the first study in Pakistan and one of the few in South Asia to examine the determinants of crop choice and soil and water conservation practices individually and jointly. It assesses the impact of these practices on crop revenue and risk exposure in mixed-cropping zones.

Notably, our study employs parametric and non-parametric methods to thoroughly examine the determinants of SLU practices, their impact on farm performance, and land use sustainability. Additionally, by comparing the three sustainability dimensions—ecological, social, and economic—we aim to provide a comprehensive land use sustainability strategy that addresses environmental, environmental and economic aspects of agricultural sustainability for both adopters and non-adopters. This study’s selection of sustainability indicators was guided by their relevance to the multi-dimensional nature of SLU practices and their ability to capture the environmental, economic, and social aspects of sustainability.

2. Conceptual Framework and Econometric Specifications

To examine the determinants of SLU practices adoption and their impacts on farm performance and risk exposure, we follow the approaches outlined in studies by [24,28] to compute crop revenue and skewness, using the third central moment of crop revenue. Skewness of crop revenue serves as an effective indicator of farm performance, particularly under climate variability, as it captures downside risk exposure. An increase in crop revenue skewness indicates a reduced probability of crop revenue loss or crop failure [28].

To estimate the crop revenue moments, we use a sequential approach. First, we regress per-acre crop revenue on inputs and related farm variables, from which residuals are retrieved. Second, we compute the third moment, following the methodology of [28]. The third moment of crop revenue is then used as the outcome variable in the MESR model to assess the impact of individual or joint SLU practices on downside risk exposure.

Modeling the Choice of SLU Practices Adoption

We develop a latent utility model to analyze farmers’ adoption decisions for SLU practices. In this framework, $V_{ij}^{*}$ represents a farmer’s unobserved (latent) utility from adopting a specific climate-smart practice J. Building on this theoretical foundation, we formalize the adoption decision process through the following expression:

$$V_{ij}^{*}=X_{ij}{\beta }_j+{\eta }_{ij}$$

(1)

In this framework (Equation (1)), the vector $X_{ij}$ captures socio-economic, institutional, and farm-level characteristics influencing adoption decisions, while ${\eta }_{ij}$ denotes the stochastic error term. The adoption behavior is modeled using a multinomial logit specification, categorizing farmers into three mutually exclusive groups: non-adopters (no SLU practices adopted), single-practice adopters (e.g., water or soil conservation practices), and joint adopters (combined adoption of water and soil conservation practices).

Within the multinomial logit framework, the probability $P_{ij}$ that a farmer selects a specific SLU practice J from M available options is defined as follows:

$$P_{ij}={\displaystyle \frac{\exp \left(X_{ij}{\beta }_j\right)}{{\sum }_{K=1}^{M}exp\left(X_{ik}{\beta }_k\right)}}$$

(2)

This modeling enables the classification of adoption decisions into discrete categories: non-adoption, adoption of water conservation practices only, adoption of soil conservation practices only, or joint adoption of both practices.

3. Data and Methods

3.1. Data

The data for this study were collected through a field survey conducted during the 2024/2025 crop season across eight districts and three agro-ecological zones in Punjab, Pakistan (Figure 1). A multistage sampling approach was employed to select and interview 506 farm households across maize-wheat, rice-wheat, and cotton-mixed zones. To begin, two tehsils were randomly chosen from each district, followed by the random selection of three union councils from each tehsil. Within each village, 8–12 households were randomly selected in proportion to the number of farmers. This resulted in 506 interviews, representing three agro-ecological zones: 152 households from the maize-wheat zone, 175 from the rice-wheat zone, and 179 from the cotton-mixed zone. The primary purpose of selecting farmers from the three agro-ecological zones was to compare the adoption of SLU practices and their resulting impacts across different climate, crop, and market conditions.

To measure land use sustainability, we compared the sustainability of land use practices between joint adopters (132) and non-adopters (92). The three dimensions of land use sustainability—ecological, social, and economic—were assessed using the [29] framework. Table 1 outlines the dimensions, indicators, and measures used for sustainability. The ecological dimension examines indicators such as integrated pest management, land environmental quality, and the consumption of nitrogen (N), potash (P), phosphorus (K), fungicides, and pesticides. The social dimension includes indicators like the number of household members working on the farm, income sources (both farm and non-farm), job satisfaction as a farmer, sustainable agricultural knowledge, and access to extension services. The economic dimension focuses on access to farm credit, crop insurance, farm size, crop yield, and input use sustainability for both joint adopters and non-adopters.

As previously discussed, SLU practices include crop choice and the adoption of soil and water conservation measures. Descriptive statistics reveal that 24.2% of farmers have adopted crop choice practices, while 31.4% have implemented soil and water conservation measures (Table 2). This indicates that one-third of the farmers are practicing techniques such as raised-bed planting, minimum tillage, zero-tillage rice sowing, organic manure application, and laser land leveling. Furthermore, 26.1% of farmers are jointly adopting both crop choice and soil and water conservation practices, while 18.3% are non-adopters.

Table 2 provides descriptive statistics for all study variables. Given that our sample includes farmers from three agro-ecological zones, the crop revenue variable was constructed by summing the per-acre crop value for one year, following the method outlined by [30]. On average, farmers earned crop revenue of 163,294 rupees per acre per year, typically comprising 2–3 crops. In addition, we also collected information on various variables, including the types of crops cultivated, socio-demographic factors, and indicators of information and knowledge, such as access to climate-related information. We recorded farmers’ perceptions of drought and heatwaves, as well as the current practices they are implementing to cope with these climate events.

To ensure our empirical estimations’ robustness and statistical reliability, we also gathered data on several control variables and covariates. This included household characteristics such as the age, gender, and education level of the household head, farm input usage (e.g., herbicides, fertilizers), and resource endowments (e.g., farm size, livestock count). Additionally, we recorded information on land ownership status (whether the land was owned or leased). These variables were selected based on insights from key studies on technology adoption, SLU practices adoption, and impact assessments [12,24,28]. To address potential endogeneity issues related to off-farm income and access to farm advisory services, both of which can influence the adoption of SLU practices, we employed two instrumental variables (IVs). The first instrument is the distance to the tehsil, which helps control for potential endogeneity stemming from access to off-farm work. The second instrument is the distance to the nearest public extension office. Given that SLU practices are labor-intensive, participation in off-farm work could be potentially endogenous, as off-farm employment may negatively affect household adoption of SLU practices due to time and engagement constraints. Additionally, off-farm income could influence the purchase of inputs, affecting investments in SLU practices due to income effects. Similarly, access to farm advisory services may positively influence SLU adoption. Both instruments satisfy the key criteria of exclusion restriction, validity (exogeneity), and relevance. The exclusion restriction refers to the condition that an instrument must fulfill to ensure that it is valid for use in the model. Hence, it reinstates that these IVs are not directly related to the dependent variable but are significantly associated with the endogenous explanatory variables.

Table 1. Land use sustainability dimensions and indicators with measurement methods.

Dimension	Indicator [Reference]	Explanation
Ecological dimension	Ecological quality of land [31,32,33,34]	It was assessed using characteristics such as soil salinity, soil texture, weed growth rate, crop growth rate, and soil color.
	Integrated pest management [31]	This indicator is assessed through methods such as natural methods, pest-resistant cultivars, crop rotation, burning infected plant parts, and seed disinfection.
	Ecological farm management [31,35]	This index includes practices such as second cultivation, mixed cropping, fallowing, alternating crops, continuous cultivation of a single crop, and the use of animal dung and organic fertilizers.
	Reverse NPK fertilizer consumption per hectare [33,35,36]	The amount of NPK fertilizer consumption for crop cultivation (kg/ha).
	Reverse insecticide, pesticide, and fungicide consumption per hectare [33,35,36]	The amounts of insecticides, pesticides, and fungicides consumed for crop cultivation (liters/hectare).
Social dimension	Multiple activities (non- agricultural income) [34,36]	This index reflects the significance of non-agricultural activities in rural areas, based on responses to an open-ended question about income from off-farm activities.
	Family workforce [36,37]	Family members are working as laborers at the farm.
	Sustainable agricultural knowledge [31,32,33,34]	Sustainability knowledge encompasses an understanding of sustainable agriculture practices. This variable is measured using 27 questions across crop rotation, tillage management, plant residue management, integrated pest and weed management, plant diseases, and water resource management.
	Job satisfaction [37,38,39]	Satisfaction with agricultural activities, income, partners, career prospects, and promotions from officers and centers, as measured by the survey.
	Possessing technical-promotional services [36,39]	The farm plot is evaluated based on class participation, educational activities, school farms, and programs.
	Access to credits [31,32,33,34]	The average number and amount of loans received measure this indicator.
	Insurance coverage [36,37]	This indicator is measured by calculating the land under crop insurance to total land.
Economic dimension	The mechanization operations [36,37]	The measurement was based on the use or non-use of agricultural machinery, including tractors, disks, plows, seeders, fertilizer machines, sprayers, and combines, within the unit area.
	Land ownership [36,37]	The total amount of agricultural land owned by the farmer.
	Farm size (reversed) [36,37]	The total number of hectares owned by the farmer.
	The size of the pieces [36,37]	The average size of each piece of land owned by the farmer.
	Agricultural income [31,32,33,34]	The total annual farm income received by a farmer from farm activities.
	Yield [34,36,37]	The crop yield of two major crops—wheat and rice—is harvested (tons/hectare).
	Input Productivity [34,36,37]	This indicator is measured by quantifying production (tons/hectare) for three inputs: fertilizers, pesticides, and seeds per hectare.

Table 1. Land use sustainability dimensions and indicators with measurement methods.

Dimension	Indicator [Reference]	Explanation
Ecological dimension	Ecological quality of land [31,32,33,34]	It was assessed using characteristics such as soil salinity, soil texture, weed growth rate, crop growth rate, and soil color.
	Integrated pest management [31]	This indicator is assessed through methods such as natural methods, pest-resistant cultivars, crop rotation, burning infected plant parts, and seed disinfection.
	Ecological farm management [31,35]	This index includes practices such as second cultivation, mixed cropping, fallowing, alternating crops, continuous cultivation of a single crop, and the use of animal dung and organic fertilizers.
	Reverse NPK fertilizer consumption per hectare [33,35,36]	The amount of NPK fertilizer consumption for crop cultivation (kg/ha).
	Reverse insecticide, pesticide, and fungicide consumption per hectare [33,35,36]	The amounts of insecticides, pesticides, and fungicides consumed for crop cultivation (liters/hectare).
Social dimension	Multiple activities (non- agricultural income) [34,36]	This index reflects the significance of non-agricultural activities in rural areas, based on responses to an open-ended question about income from off-farm activities.
	Family workforce [36,37]	Family members are working as laborers at the farm.
	Sustainable agricultural knowledge [31,32,33,34]	Sustainability knowledge encompasses an understanding of sustainable agriculture practices. This variable is measured using 27 questions across crop rotation, tillage management, plant residue management, integrated pest and weed management, plant diseases, and water resource management.
	Job satisfaction [37,38,39]	Satisfaction with agricultural activities, income, partners, career prospects, and promotions from officers and centers, as measured by the survey.
	Possessing technical-promotional services [36,39]	The farm plot is evaluated based on class participation, educational activities, school farms, and programs.
	Access to credits [31,32,33,34]	The average number and amount of loans received measure this indicator.
	Insurance coverage [36,37]	This indicator is measured by calculating the land under crop insurance to total land.
Economic dimension	The mechanization operations [36,37]	The measurement was based on the use or non-use of agricultural machinery, including tractors, disks, plows, seeders, fertilizer machines, sprayers, and combines, within the unit area.
	Land ownership [36,37]	The total amount of agricultural land owned by the farmer.
	Farm size (reversed) [36,37]	The total number of hectares owned by the farmer.
	The size of the pieces [36,37]	The average size of each piece of land owned by the farmer.
	Agricultural income [31,32,33,34]	The total annual farm income received by a farmer from farm activities.
	Yield [34,36,37]	The crop yield of two major crops—wheat and rice—is harvested (tons/hectare).
	Input Productivity [34,36,37]	This indicator is measured by quantifying production (tons/hectare) for three inputs: fertilizers, pesticides, and seeds per hectare.

Appendix A,

Table A1. Test the validity of the instruments used to identify the MESR model.

Variables	Crop Revenue of Non-Adopters	Revenue Skewness of Non-Adopters
Drought-perception	0.156 (0.197)	−0.314 (0.381)
Heatwave-perception	−0.183 (0.159)	0.248 (0.264)
Farm advisory	0.156 (0.167)	0.521 (0.516)
Climate-info	−0.073 (0.116)	0.183 (0.232)
Distance-tehsil	0.042 (0.155)	0.273 (0.155)
Constant	3.135 *** (0.541)	2.471 *** (0.612)
F-test on instruments	2.329 [p = 0.416]	1.619 [p = 0.311]

Note: Standard errors in parentheses. p-values of the F-test in square brackets represent the validity of instruments. *** indicates 1% significance levels.

and

Table A2. Parameter estimate of risk exposure by SLU practices: second-stage MESR results.

Variables	Non-Adopter (n = 92)		Crop Choice (n = 122)		Soil and Water Cons (n = 158)		Joint Adoption (n = 132)
	Coeff.	SE	Coeff.	SE	Coeff.	SE	Coeff.	SE
Constant	0.738	1.12	1.17	1.32	0.62	0.58	0.76	0.71
Age	−0.501	0.41	−1.09 ***	0.23	−0.58	0.41	1.41	1.61
Gender	−1.82 ***	0.52	0.82	0.65	−0.57	1.45	−1.08	0.98
Off-farm	3.57 **	1.72	−1.27 **	0.54	1.04 ***	0.32	3.27 ***	0.87
Education	−0.08 ***	0.02	0.89 **	0.41	2.74 ***	0.87	3.15 ***	0.57
Experience	−0.72	0.82	−1.61 **	0.71	0.39	0.43	−0.91	0.67
Farm advisory	−0.92	1.02	1.05 ***	0.35	−2.31 ***	0.57	4.05 ***	0.96
Livestock	0.71	0.83	0.15	0.27	2.38 ***	0.68	−3.50	2.45
Farm size	−0.52 ***	0.09	−0.85	0.65	−3.74	2.96	−1.26	0.78
Hired labor	1.62 ***	0.51	−2.14 ***	0.52	1.05	1.18	0.89	1.13
Extension	1.03	1.22	3.91 ***	0.67	−0.77 ***	0.22	3.52	2.85
Landowner	−0.77	0.72	−0.43	1.12	0.76	0.89	−0.85	0.73
Tenant	−0.52	0.48	−0.63	0.53	−1.24	1.62	−0.76	1.02
Fertilizer	−2.31 **	0.92	1.15 ***	0.42	1.82	1.52	1.20	1.27
Pesticides and herbicides	−0.06	0.05	−1.17	0.84	1.06	1.15	−0.92	1.32
Climate info	0.74	1.61	2.13	1.81	−1.17	1.43	0.82 **	0.41
Selectivity terms
m1	−0.08 ***	0.01	0.17	0.26	0.70	0.76	0.93	0.85
m2	0.22	0.31	−0.75 **	0.34	−0.46	0.55	0.83 ***	0.27
m3	−0.45	0.46	0.08	0.06	0.61 *	0.32	−1.90	2.16
m4	0.52	0.48	0.32	0.45	0.77	0.85	0.90	0.89

Note: Dependent variable: crop revenue skewness. ***, **, * indicate 1%, 5%, and 10% significance levels, respectively. SE represents bootstrapped standard errors.

report the means of various covariates related to choices. The data indicate significant differences in crop revenues among the options. However, these differences do not account for selection bias arising from both observed and unobserved factors. This suggests that these covariates may impact farm performance differently depending on the adoption of SLU practices. Therefore, it reinforces the need for the use of the MESR model in this study.

Table 2. Descriptive statistics of farm household variables regarding the SLU practices adoption.

Variables	Variable Description	Mean	SD
Crop revenue	Total crop revenue (PKR †/acre)	163,294.4	71,752.3
Farming experience	Number of years in farming	23.27	12.74
Fertilizer	Expenditures on organic and inorganic fertilizers (PKR/acre)	17,328.3	92,715.2
Pesticides and herbicides	Expenditures on organic and inorganic fertilizers (PKR/acre)	5820.6	3863.5
Farm size	Number of acres	9.72	7.84
Hired labor	Expenditures on hired labor (PKR/acre)	11,486.27	7211.4
Education	Year of formal education	9.73	6.39
Age	Age of the farmer in years	39.58	14.22
Gender	Male = 1; female 0	97.28	0.23
Off-farm	Farmer engaged in off-farm work = 1; else 0	0.39	0.45
Farm advisory	Number of extensions visits per crop	1.21	0.83
Livestock	Livestock ownership in tropical livestock units (TLU ‡)	2.74	2.89
FBO	FBO members = 1; else 0	0.26	0.43
Distance-city	Distance to the tehsil	4.35	3.96
Distance-Ext.	Distance to nearest extension office	2.81	1.45
Perception-drought	Perception of drought occurrence = 1, else 0	0.77	0.42
Perception-heatwave	Perception of heatwave occurrence = 1, else 0	0.81	0.33
Landowner	Farmer is also landowner = 1; else 0	0.43	0.53
Tenant	Farmer is tenant = 1; else 0	0.34	0.26
Climate-info	Farmer receive climate information = 1; else 0	0.59	0.46
Crop choice	Percentage of farmers practicing crop choice	24.2	-
Soil and water cons	Percentage of farmers practicing soil and water conservation	31.4	-
Joint	Percentage of farmers practicing both	26.1	-
Non-adoption	Percentage of farmers with no adoption	18.3	-
N		504

Note: ^† Exchange rate at the time of survey was 1 USD = 281.3 PKR. ^‡ TLU conversion factors are sheep = 0.1, cattle and buffalo = 0.7, goats = 0.1. SD refers to standard deviation.

Table 2. Descriptive statistics of farm household variables regarding the SLU practices adoption.

Variables	Variable Description	Mean	SD
Crop revenue	Total crop revenue (PKR †/acre)	163,294.4	71,752.3
Farming experience	Number of years in farming	23.27	12.74
Fertilizer	Expenditures on organic and inorganic fertilizers (PKR/acre)	17,328.3	92,715.2
Pesticides and herbicides	Expenditures on organic and inorganic fertilizers (PKR/acre)	5820.6	3863.5
Farm size	Number of acres	9.72	7.84
Hired labor	Expenditures on hired labor (PKR/acre)	11,486.27	7211.4
Education	Year of formal education	9.73	6.39
Age	Age of the farmer in years	39.58	14.22
Gender	Male = 1; female 0	97.28	0.23
Off-farm	Farmer engaged in off-farm work = 1; else 0	0.39	0.45
Farm advisory	Number of extensions visits per crop	1.21	0.83
Livestock	Livestock ownership in tropical livestock units (TLU ‡)	2.74	2.89
FBO	FBO members = 1; else 0	0.26	0.43
Distance-city	Distance to the tehsil	4.35	3.96
Distance-Ext.	Distance to nearest extension office	2.81	1.45
Perception-drought	Perception of drought occurrence = 1, else 0	0.77	0.42
Perception-heatwave	Perception of heatwave occurrence = 1, else 0	0.81	0.33
Landowner	Farmer is also landowner = 1; else 0	0.43	0.53
Tenant	Farmer is tenant = 1; else 0	0.34	0.26
Climate-info	Farmer receive climate information = 1; else 0	0.59	0.46
Crop choice	Percentage of farmers practicing crop choice	24.2	-
Soil and water cons	Percentage of farmers practicing soil and water conservation	31.4	-
Joint	Percentage of farmers practicing both	26.1	-
Non-adoption	Percentage of farmers with no adoption	18.3	-
N		504

Note: ^† Exchange rate at the time of survey was 1 USD = 281.3 PKR. ^‡ TLU conversion factors are sheep = 0.1, cattle and buffalo = 0.7, goats = 0.1. SD refers to standard deviation.

3.2. Multinomial Endogenous Switching Regression Model

The MESR approach has been employed in studies examining SLU adoption and agricultural technology adoption at the farm level [12,23,28]. The MESR model is preferable to the Heckman two-stage model because it can handle multiple outcome categories, especially when dealing with multistage decision processes and endogeneity issues. It is highly flexible in accounting for heterogeneous effects and provides more consistent and accurate estimates in complex scenarios. Importantly, the MESR model is better suited for analyzing data with non-binary outcomes and multiple decision-making stages. Given more than two outcome stages in our study, the MESR model is the most appropriate choice. This approach treats non-adoption as the base category (J = 1), with crop choice as J = 2, adoption of soil and water conservation practices as J = 3, and joint adoption of both practices as J = 4.

The observed adoption outcomes allow us to construct an outcome equation (or farm performance model). The impact of adopting SLU practices on crop revenue is modeled as follows:

$$Y_{ij}=Z_{ij}{\alpha }_j+{\rho }_j{\lambda }_{ij}+{ϵ}_{ij}$$

(3)

In this framework (Equation (3)), $Z_{ij}$ denotes key explanatory variables (e.g., access to farm advisory services and inputs), ${\lambda }_{ij}$ represents the inverse Mills ratio derived from Equation (2) to correct for selection bias, and ${\rho }_j$ captures the correlation between unobserved factors influencing adopting climate-smart practices and their subsequent impact on farm revenue.

To model risk exposure—particularly downside risk—we approximate it by calculating the third central moment (skewness) of the distribution as follows:

$${Skew}_{ij}={\gamma }_0+Z_{ij}{\gamma }_j+v_{ij}$$

(4)

In Equation (4), the expression ${Skew}_{ij}$ quantifies the skewness of farm performance (crop revenue) by analyzing residuals derived from a crop-revenue regression model. A higher skewness value indicates lower downside risk exposure, reflecting a reduced probability of severe crop revenue losses.

Building on this approach, we estimate the treatment effect to quantify the causal impact of adopting SLU practices on the outcomes of interest.

$${ATT}_j=E[Y_j\mid D_j=1]-E[Y_{i0}\mid D_j=1]$$

(5)

The ATT (Equation (5)) quantifies the difference between the observed outcomes for adopters of a specific SLU practice J and the counterfactual outcomes (i.e., what would have occurred if they had not adopted SLU practices).

Given the potential endogeneity of farmers’ access to farm advisory services and off-farm income, we employ a control function approach to address this bias, resulting in the following expressions:

$$D_i=W_i\delta +u_i (\mathrm{first} \mathrm{stage})$$

(6)

$$Y_{ij}=Z_{ij}{\alpha }_j+{\theta \widehat{u}}_i+{ϵ}_{ij} (\mathrm{second} \mathrm{stage})$$

(7)

In Equation (6), the term $D_i$ represents the endogenous variable (e.g., off-farm income), which may be binary or continuous. The term $W_i$ includes exogenous covariates and instrumental variables. For instance, distance to the city may influence the likelihood of engaging in off-farm work, but it is unlikely to directly affect crop revenue. The error term $u_i$ captures unobserved factors influencing the endogenous regressor.

In Equation (7), we use the residual ${\widehat{u}}_i$, derived from the first stage, in the second-stage outcome equation to estimate farm revenue and downside risk exposure. Including this residual accounts for selection bias and yields more consistent treatment effect estimates. In this setup, $Y_{ij}$ denotes the outcome variables (e.g., farm performance or risk exposure), $Z_{ij}$ represents the explanatory variables (such as farm size, fertilizer use, farmer’s age, etc.), and ${\widehat{u}}_i$ is the control function term. A statistically significant ${\widehat{u}}_i$ confirms the presence of endogeneity in Equation (6).

Building on this framework, we estimate joint adoption’s complementarity (modeled as an interaction term) by testing for synergistic effects between water and soil conservation practices.

$$Y_i={\delta }_0{+\delta }_1{D}_A{+\delta }_2{D}_B{+\delta }_3\left(D_A\times D_B\right)+X_i\beta +{\mathsf{ϵ}}_i$$

(8)

In this framework, $D_A$ and $D_B$ are dummy variables indicating the adoption of water and soil conservation practices, respectively. The interaction term $\left(D_A\times D_B\right)$ evaluates whether these practices exhibit substitutability (competing effects) or complementarity (synergistic effects) within the study area’s SLU adoption at the farm level.

3.3. Robustness Checks

To ensure the validity of our estimations and, therefore, the reliability of the findings, we conducted robustness checks to test the validity of the instruments. First, we applied a falsification test of the instruments, following the methodology of [12]. Second, we performed robustness checks on the study results by using an alternative method—multivariate treatment effects. This approach better addresses endogeneity arising from unobserved factors associated with multinomial choices [40].

To address these potential endogeneity issues, we used a control function approach [41]. In this specification, we first utilized two instruments in the first-stage probit model (distance to tehsil and distance to nearest public extension office), then employed the predicted values of access to farm advisory services and off-farm work in the second stage. The observed values of the endogenous variables and the generalized residuals from the first stage were included as covariates in the MESR model. These residuals function as part of the control function approach and help to provide consistent estimates [41].

3.4. Land Use Sustainability Measurements

To measure sustainability, we first standardized the indicators by identifying the minimum and maximum values for each, then calculating the range of variation. Next, we subtracted the minimum value from the indicator value, divided the result by the range of variation, and scaled it to fall between 0 and 1, following the method outlined by [29].

$$r_{ij}={\displaystyle \frac{X_{ij}-X_j^{min}}{X_j^{max}-X_j^{min}}}$$

(9)

where $r_{ij}$ represents the value of indicator i, term $X_j^{min}$ denotes the minimum value of indicator i, and $X_j^{max}$ indicates the maximum value of indicator i. We used Cronbach’s alpha (CA) to test the validity of the measures of the three sustainability dimensions (Appendix A 

Table A3. Robustness checks: Multivariate treatment effect regression ^†.

SLU Practice	Coefficient	SE
Crop revenue (log)
Crop choice	0.417 ***	0.072
Soil and water cons	0.289 ***	0.045
Joint adoption	0.627 ***	0.107
Risk exposure (skewness/downside risk)
Crop choice	0.419 ***	0.053
Soil and water cons	0.272 ***	0.021
Joint adoption	0.716 ***	0.009

Note: *** indicates 1% significance levels. ^† Reference category is non-adoption.

). The results confirm the validity of all constructs, since CA for all dimensions was above the threshold level (0.75).

4. Results and Discussion

4.1. Robustness Checks Result

The results of the determinants of SLU practices adoption are presented in

Table 3. Parameter estimates of SLU practices adoption: multinomial logit selection model ^†.

Variables	Crop Choice (n = 122)		Soil and Water Cons (n = 158)		Joint Adoption (n = 132)
	Coeff.	SE	Coeff.	SE	Coeff.	SE
Constant	−16.38	13.59	−11.75	9.831	−22.41	17.24
Age	−1.274 *	0.682	−0.713 **	0.327	−1.639 **	0.817
Gender	0.527	0.373	1.225	1.172	0.715	0.548
Education	1.341 ***	0.068	2.028 ***	0.214	3.408 ***	0.672
Off-farm	−2.639 ***	0.652	−1.294 *	0.394	−3.259 ***	0.778
Farming experience	3.072	2.705	0.315 ***	0.087	1.282	1.056
Farm advisory	1.825 **	0.822	2.432 **	0.652	4.816 ***	1.511
Livestock	1.386	1.146	2.657 ***	0.601	3.271 ***	0.719
FBO	2.911 ***	0.361	4.836	3.514	2.478	1.892
Farm size	5.846 ***	1.138	1.371	0.987	0.763	0.606
Hired labor	0.535 *	0.272	0.835 ***	0.162	1.356 ***	0.325
Distance-city	−3.794 ***	0.742	−2.722 ***	0.623	−2.038 **	0.823
Distance-Ext.	−1.837 ***	0.624	3.015	1.915	−2.177	1.812
Perception-drought	5.921 ***	1.912	4.031 ***	0.863	6.581 ***	2.379
Perception-heatwave	7.385 ***	2.081	5.867 **	1.238	4.924 ***	1.081
Landowner	2.337 **	1.032	0.682	0.611	2.381 **	1.157
Tenant	−0.594	0.428	3.011	1.839	−1.013	0.931
Climate-info	2.193 ***	0.285	3.314 ***	0.615	6.826 **	2.426
Fertilizer	3.982	2.745	1.267 ***	0.427	2.063 ***	0.635
Pesticides and herbicides	0.396	0.285	3.609 ***	0.722	1.912	1.661
Resid-Off-farm	−1.842	1.635	0.623	0.503	0.521	0.485
Resid-Extension	0.371	0.281	0.538	0.397	0.417	0.334
Joint sig instruments (χ*) in crop revenue equation	62.32 ***		43.67 ***		25.31 ***
Joint sig instruments (χ*) in skewness equation	79.27 ***		35.19 ***		17.21 ***
Skewness equation Wald test, χ*	253.16
N	504

Note: ***, **, * indicate 1%, 5%, and 10% significance levels, respectively. ^† Reference category is non-adoption.

, with non-adoption serving as the reference category. The significance of the Wald test (χ* = 253.16, p > χ* = 0.00) rejects the null hypothesis. Thus, it confirms that the regression coefficients are jointly non-zero. This result also validates the joint significance of the instruments—drought perception, heatwave perception, climate information, and FBO membership—used to identify the MESR model.

The falsification test results further support the validity of the instruments (see Appendix A 

Table A4. Cronbach’s alpha results for dimensions of sustainability.

Factors	Number of Questions	Cronbach’s Alpha
Ecological quality of land	13	0.817
Integrated pest management	8	0.793
Job satisfaction	12	0.825
Possessing technical-extensional services	7	0.832

). While the excluded instruments show a joint impact on adoption at all levels, they do not affect risk exposure or crop revenue. These findings highlight the influence of household characteristics, climate-related information, farmers’ endowments, and farm-specific covariates on the adoption of crop choice, soil and water conservation practices, and the joint adoption of both (Table 3). Notably, the education level of the farm household head has a significant and positive impact on the adoption of individual practices (crop choice and soil and water conservation) as well as their joint adoption. These results align with recent studies and emphasize the importance of considering farmer-specific characteristics when designing and implementing strategies to promote the broader adoption of SLU practices at the farm level [42,43].

The coefficient for age is significantly negative across all practices, indicating that older farmers tend to adopt fewer SLU practices at the farm level. This aligns with previous studies suggesting that older farmers are generally less inclined to adopt new technologies [44,45], including SLU practices. The education level of the household head has a significant positive impact on the adoption of all SLU practices. This suggests that as the education level of farmers increases, their willingness to adopt SLU practices also rises, which is consistent with findings from earlier research. Importantly, the coefficient for off-farm income is significantly negative for adopting crop choice, soil and water conservation, and the joint adoption of both. This highlights the labor-intensive nature of SLU practices, suggesting that farmers engaged in off-farm work have less time and capacity to adopt SLU practices. This finding is in line with expectations and prior studies as well [46,47], given the time and labor demands associated with SLU practices. The coefficient for farming experience is positively related to adopting soil and water conservation practices. This supports previous studies that emphasize the importance of knowledge in promoting the uptake of SLU practices [15,47]. The findings show that access to farm advisory services significantly and positively impacts adopting all practices. This reinforces the critical role that education and advisory services play in promoting SLU adoption at the farm level. These results are in line with [15,48], which underscores the importance of farm advisory services in encouraging SLU adoption.

The coefficient for livestock ownership (TLU) is significant and positive for the adoption of soil and water conservation practices and for the joint adoption of both practices. This suggests that resource-endowed farmers are more likely to engage in SLU practices. These findings reinforce the idea that ownership of resources enables farmers to hire labor and utilize organic manure and other essential inputs required for SLU adoption at the farm level. This aligns with [46], which emphasizes the crucial role of farmers’ resource endowments in technology and SLU adoption. The coefficient for FBO membership and farm size is significant and positive for crop choice adoption. Similarly, the coefficient for hired labor is significant for all choices, indicating that farmers who employ hired labor are more likely to adopt SLU practices, given the labor-intensive nature of SLU, such as manual weed control. In contrast, commercial farms typically rely on herbicides and other chemicals, which require less labor and time. Interestingly, the coefficient for distance to tehsil is negative for all choices, implying that farmers in more remote areas tend to adopt fewer SLU practices. This is likely due to the higher costs and, in some cases, lower economic returns associated with these practices, particularly when farmers lack the necessary skills. Remote farmers face additional transportation costs to move their climate-efficient products to cities, further increasing their production costs and discouraging adoption.

Notably, the coefficients for drought and heatwave perceptions are positive and significant for all practices. This indicates that farmers’ perceptions of climate events influence their adoption behavior. Farmers who have experienced adverse climate events or observed changing patterns in rainfall and temperature are more likely to adopt mitigation strategies at the farm level. These findings are consistent with previous studies [49,50,51], which highlights the critical role of farmers’ perceptions in technology adoption.

The results indicate that the coefficient of climate information is significant and positive for all choices. It emphasizes the crucial role of knowledge and access to up-to-date information in the broader adoption of SLU practices. This suggests that SLU practices require expert advice and relevant climate change information to help farmers understand the severity of climate change impacts and the economic and ecological benefits of adopting SLU practices to mitigate these adverse effects at the farm level. The coefficient for fertilizer is also significant and positive for the adoption of soil and water conservation practices and for the joint adoption of both practices at the farm level. This highlights that using fertilizers and herbicides is a common practice among farmers to reduce labor costs, supporting the efficiency of SLU practices implementation. As previously mentioned, to address the potential endogeneity of farmers’ participation in off-farm work and access to farm advisory services, we employed a control function approach and estimated the residual terms (Resid-Off-farm, Resid-Extension) from the first stage of the CF regression. The results show that neither variable is significant in any of the choices. Thus, it confirms the exogeneity of off-farm work and access to farm advisory services concerning the adoption of SLU practices.

4.2. Crop Revenue and Downside Risk Exposure: Second-Stage MESR Estimates

We estimated the determinants of crop revenue and downside risk exposure (skewness) based on farmers’ choice of SLU practices. We included four selectivity terms (m1, m2, m3, and m4) to account for selectivity arising from unobserved characteristics. To address heteroskedasticity, we employed a bootstrapping process with 100 replications for the estimated variances, following [27]. In the revenue equation, the selectivity correction terms are significant for the non-adoption category, soil and water conservation, and joint adoption of both practices. This confirms the presence of sample selectivity, indicating that OLS estimates would be biased and inconsistent. Using the MESR approach, we accounted for selectivity and heteroskedasticity, providing unbiased and consistent estimates.

The results show that the use of chemical fertilizer significantly influences crop revenue among adopters of soil and water conservation practices, as well as those adopting both practices (

Table 4. Parameter estimate of crop revenue (log) by SLU practices adoption: second-stage MESR results.

Variables	Non-Adopter (n = 92)		Crop Choice (n = 122)		Soil & Water Cons (n = 158)		Joint Adoption (n = 132)
	Coeff.	SE	Coeff.	SE	Coeff.	SE	Coeff.	SE
Constant	−2.05	1.79	−1.373	1.422	−0.637	0.536	−1.523	1.317
Age	0.741 **	0.352	−1.416	1.552	−0.522	0.472	0.632	0.501
Gender	−2.272 ***	0.563	0.354	0.416	1.534	0.981	0.294	0.217
Off-farm	−1.934 ***	0.472	2.321 ***	0.627	1.063 ***	0.363	1.482 **	0.673
Education	1.871 ***	0.381	1.336	0.936	2.282 ***	0.732	1.047 *	0.556
Experience	2.253 ***	0.931	0.756 **	0.344	0.411 ***	0.062	0.761 **	0.351
Farm advisory	1.514	1.032	1.483	1.341	2.326 ***	0.642	0.992 **	0.465
Livestock	1.376	1.054	0.592	0.445	2.393	2.021	1.683	1.348
Farm size	−0.115	0.161	1.547 ***	0.523	−3.757	2.862	−0.762	0.677
Hired labor	2.234	1.633	0.354	0.257	1.065	0.975	1.084 **	0.463
Extension	2.438 ***	0.824	2.029	1.363	0.791 **	0.392	2.026	1.944
Landowner	0.317	0.213	2.034	1.731	0.772	0.647	1.146	0.972
Tenant	−0.148	0.092	0.543	0.628	−1.252	0.935	1.763	1.346
Fertilizer	1.537 **	0.731	2.017	1.815	1.836	1.356	1.572 **	0.693
Pesticides and herbicides	0.753 **	0.325	0.543 ***	0.047	1.073 ***	0.227	1.116 ***	0.432
Climate info	2.038	1.892	1.371 ***	0.073	1.185 ***	0.301	2.824 ***	0.537
Selectivity terms
m1	−0.943	0.683	0.363	0.322	−0.726 **	0.301	0.356 ***	0.092
m2	1.037 *	0.524	−0.517	0.481	0.471 *	0.247	0.726	0.651
m3	−0.676 **	0.282	0.324	0.302	0.623	0.526	−0.915	0.715
m4	−0.642	0.621	0.829	0.675	−0.782	0.644	0.297	0.232

Note: ***, **, * indicate 1%, 5%, and 10% significance levels, respectively. SE represents bootstrapped standard errors.

). Similarly, the coefficient for herbicides and pesticides is significantly and positively related to crop revenue for adopters of crop choice and soil and water conservation practices and joint adoption of both. These findings suggest that fertilizers and pesticides act as complementary inputs that improve crop revenue and encourage the adoption of SLU practices. The coefficient for off-farm income is significant and positively influences crop revenue for adopters of crop choice, soil and water conservation practices, and joint adoption of both. This implies that, given the labor-intensive nature of these practices, off-farm income is used to hire labor, significantly improving crop revenue. This further emphasizes the role of income effects in enhancing crop revenue. These findings are consistent with prior studies [52,53,54], highlighting the importance of off-farm income in improving farm productivity.

Notably, the coefficient for climate information significantly enhances crop revenue for adopters of crop choice and soil and water conservation practices and joint adoption of both. These results align with the FAO’s principles of SLU practices [14], which aim to enhance crop productivity, revenue, resilience, and farmers’ adaptability to climate change mitigation. The determinants of downside risk exposure (skewness) by adoption of SLU practices are detailed in Appendix A, Table A2.

4.3. Impact of SLU Practices on Crop Revenue and Downside Risk Exposure

We estimated the individual and joint impacts of SLU practice adoption on crop revenue and downside risk exposure (skewness) (

Table 5. Average treatment effect of SLU practices adoption on crop revenue and downside risk exposure.

	Adoption Decision		ATT	Outcome Change (%)	ATT by Agro-Ecological Zone
	If Adopters Had Adopted	If Adopters Had Not Adopted			Maize-Wheat	Rice-Wheat	Cotton-Mixed
Crop revenue (log)
Crop choice	4.262	3.648	0.614 *** (0.123)	16.83	0.384 *** (0.047)	0.728 *** (0.142)	0.422 *** (0.083)
Soil and water cons	5.473	5.012	0.461 *** (0.061)	9.20	0.419 *** (0.083)	0.372 *** (0.029)	0.384 *** (0.075)
Joint adoption	5.908	4.535	1.373 *** (0.314)	26.23	0.451 *** (0.092)	0.931 *** (0.155)	0.516 (0.438)
Risk exposure (skewness/downside risk)
Crop choice	−1.348	−0.972	0.376 *** (0.023)	32.67	0.216 *** (0.045)	−0.264 *** (0.063)	0.209 *** (0.052)
Soil and water cons	0.774	0.517	0.257 *** (0.056)	49.71	0.153 *** (0.022)	0.162 ** (0.071)	0.317 *** (0.031)
Joint adoption	0.282	0.145	0.137 *** (0.011)	69.54	0.117 ** (0.052)	0.138 *** (0.016)	0.413 *** (0.059)

Note: ***, ** indicate 1% and 5% significance levels, respectively.

). Two scenarios were considered: the observed scenario, where farmers adopted SLU practices, and the counterfactual scenario, assuming non-adoption. The results reveal that the adoption of SLU practices leads to significant gains in crop revenue. The highest log revenue effect of 1.373 is observed with the joint adoption of SLU practices, resulting in an approximate 26% increase in crop revenue. This is followed by the adoption of crop choice, which leads to a 16% increase in crop revenue. Similarly, adopting soil and water conservation practices results in a 9.2% increase in crop revenue for adopters. These findings are consistent with studies from Sub-Saharan Africa [26], South Asia [5], and China [8], suggesting that the adoption of SLU practices significantly boosts crop revenue. In counterfactual scenarios, adopters would experience lower crop revenue if they had not adopted SLU practices.

Regarding downside risk exposure (skewness), the results indicate that the joint adoption of both SLU practices leads to the most significant reduction in the likelihood of revenue loss or crop failure. Specifically, joint adoption reduces the chances of crop failure or revenue loss by 69.5%. Adopting crop choice alone reduces the likelihood by 32.6%, while adopting soil and water conservation practices increases skewness by 49.7%. These findings demonstrate the substantial complementarity of SLU practices in safeguarding crop revenue and reducing the likelihood of crop failure. These findings align with [24], underscoring the significant role of SLU practices in lowering downside risk exposure. This, in turn, enhances farmers’ confidence in adopting SLU practices as a key strategy to mitigate the impacts of adverse climate events such as heatwaves, droughts, and frosts—events that cause significant crop losses and substantial revenue declines.

To ensure the robustness of the individual and joint ATT estimates, we disaggregated them by agro-ecological zones: wheat-maize, wheat-rice, and cotton-mixed. The results reveal that the joint adoption of SLU practices has the highest statistically significant positive impact on crop revenue in the rice-wheat zone (ATT = 0.931). However, joint adoption does not significantly impact the cotton-mixed zone. Notably, the results confirm that the joint adoption of SLU practices substantially reduces the likelihood of crop failure or revenue loss across all three zones. These zone-specific insights provide valuable information for devising site-specific policies to facilitate the broader implementation of SLU practices across Pakistan’s agro-ecological zones. Furthermore, following [40], we conducted a multivariate treatment effect regression as an additional robustness check (Appendix A Table A2). The results show significant positive individual and joint impacts of SLU practices, generally consistent with the findings from the MESR model. These findings align with previous studies [15,18,19], reinforcing the importance of SLU practices in enhancing crop revenue and reducing downside risk exposure.

In conclusion, our findings affirm the role of SLU practices—particularly crop choice and soil and water conservation practices (e.g., laser land leveling, raised bed planting, zero tillage, and dry sowing of rice)—as key strategies for mitigating production risks, especially in the face of climate change’s adverse impacts. Importantly, these results challenge the notion that farmers who adopt SLU practices experience lower yields and crop revenue [13]. Additionally, our study highlights the strong complementarity between SLU practices, demonstrating greater joint impacts on crop revenue and downside risk exposure. This study’s unique contribution lies in examining the joint effect of both practices, offering focused and conclusive insights into the role of SLU practices, their complementarity, location-specific effectiveness, and their potential in promoting resilient agriculture and sustainable farm income in rural communities.

4.4. Land Use Sustainability

To assess land use sustainability, we compared the three dimensions—ecological, social, and economic—using the indicators shown in

Table 6. Mean comparison of the ecological dimension indicators.

Indicators	Joint Adopters		Non-Adopters		t-Value
	Mean	SD	Mean	SD
Ecological quality of land	3.11	0.81	2.32	0.63	7.72 ***
Integrated pest management	2.78	0.52	1.61	0.27	4.64 ***
Ecological farm management	8.42	2.41	6.18	2.02	5.25 ***
Reverse NPK fertilizer consumption per hectare	251.15	93.43	239.2	84.52	6.02 ***
Reverse insecticide, pesticide, and fungicide consumption per hectare	25.04	17.55	24.87	13.91	0.83

Note: *** indicates 1% significance levels.

,

Table 7. Mean comparison of the social dimension indicators.

Indicators	Joint Adopters		Non-Adopters		t-Value
	Mean	SD	Mean	SD
Income (farm and non-farm)	1,238,626	533,785	104,368	519,831	7.43 ***
Family workforce	3.26	0.98	3.57	0.83	0.73
Sustainable agricultural knowledge	17.9	6.83	13.6	5.11	5.82 ***
Job satisfaction	3.87	1.16	3.72	0.94	0.83
Possessing technical promotional services	2.54	1.17	2.05	0.72	4.15 ***

Note: *** indicates 1% significance levels.

,

Table 8. Mean comparison of the economic dimension indicators.

Indicators	Joint Adopters		Non-Adopters		t-Value
	Mean	SD	Mean	SD
The mechanization operations	9.27	3.13	7.31	2.54	6.35 ***
Total farm size	4.53	1.05	4.65	1.12	0.65
Farm size under ownership (reversed)	3.29	0.85	3.42	0.91	0.87
Agricultural income	742,672	311,576	602,785	282,673	6.13 ***
Yield	41.6	21.4	34.9	18.4	3.82 ***
Farm credits	0.37	0.21	0.38	0.21	0.56
Crop insurance	0.13	0.07	0.14	0.08	0.76
Input Productivity	0.39	0.24	0.28	0.17	4.03 ***

Note: *** indicates 1% significance levels.

and

Table 9. Mean comparison of the overall land use sustainability indicators.

Indicators	Joint Adopters		Non-Adopters
	Sustainability Score	Status	Sustainability Score	Status
Ecological	0.81	Sustainable	0.63	Potentially Moderate (good)
Social	0.65	Potentially Moderate (good)	0.47	Moderate
Economic	0.54	Moderate	0.39	Potentially unsustainable (weak)
Total	0.67	Potentially Moderate (good)	0.49	Moderate

. An independent t-test was conducted to compare the mean values of ecological dimension indicators between joint adopters and non-adopters. Table 6 highlights significant differences in indicators such as land ecological quality, integrated pest management, ecological farm management, and reverse NPK fertilizer consumption per hectare. These findings indicate that joint adopters of land use practices demonstrate higher land use sustainability compared to non-adopters. However, the results show no significant difference between joint adopters and non-adopters’ reverse consumption of herbicides, pesticides, and fungicides. Next, we compared the social dimension indicators between joint adopters and non-adopters (Table 7). The results reveal significant differences in income sources, sustainable knowledge management, and access to technical-promotional or extension services, with joint adopters showing higher values in these areas. As mentioned earlier, considering the ecological and social dimensions, the adoption of SLU practices—particularly crop choice and soil and water conservation—can help establish effective spatial patterns of land use (including crop production, living, and ecology). These findings are in line with studies by [29,36] from Iran. This alignment with national SLU strategies can foster ecologically and socially sustainable development.

We compared the economic dimension indicators between joint adopters and non-adopters (Table 8). The results confirm that joint adopters significantly differ from non-adopters regarding mechanization operations, maize and wheat crop yields, farm income, and sustainability of input use. However, indicators such as total farm size, land ownership, access to farm credit, and crop insurance show no significant differences between the two groups. Thus, considering the economic dimension and its impact on improved mechanization, crop yields, farm income, and input use sustainability, adopting SLU practices may encourage farmers to invest in land and farm improvements. This, in turn, could enhance farm productivity, improve performance, and contribute to sustainable rural livelihoods and increased income.

Lastly, we classified the overall land use sustainability of joint adopters and non-adopters using Prescott-Allena’s five sustainability categories, following the scale suggested by [29] (Appendix A 

Table A5. Prescott-Allena five categories for classification of sustainability levels.

Status	Equivalent	Value	Rank
Unsustainable	0.0–0.2	0–20	5
Potentially unsustainable (weak)	0.2–0.4	20–40	4
Moderate	0.4–0.6	40–60	3
Potentially Sustainable (good)	0.6–0.8	60–80	2
Sustainable	0.8–1.0	80–100	1

Source: [29].

). First, we converted scale-free indicators of qualitative variables into quantitative ones, as outlined in [55]. Table 9 presents the scale-free indicators for the sustainability dimensions in both joint adopters and non-adopters. The results illustrate that joint adopters have higher ecological, social, economic, and overall sustainability scores than non-adopters. Thus, these findings reinforce that joint adoption of SLU practices improves overall land use sustainability at the farm level.

5. Conclusions and Policy Implications

This study investigates the determinants of SLU practices adoption using farm-level data from 504 households across three agro-ecological zones in Punjab, Pakistan. It explores the effects of individual and joint adoption of SLU practices on farm performance, specifically crop revenue, and exposure to downside production risk, such as crop failure. Further, using a comprehensive non-parametric approach, we assess land use sustainability across ecological, social, and economic dimensions. A key feature of our study is using data from mixed cropping systems and annual per-acre revenue, rather than monocrop revenue, which can distort the assessment of SLU impacts. To address selectivity bias from both observed and unobserved variables, we employed the newly developed MESR model. This study offers valuable insights for policy and practices.

Our results indicate that factors such as farmer education, access to farm advisory services, membership in farmer-based organizations (FBOs), hired labor, farm size, perceptions of drought and heatwaves, and access to climate information are critical determinants of SLU adoption. The study found that the joint adoption of SLU practices—specifically crop selection and soil and water conservation—yields the highest crop revenue returns. This suggests complementarity between the joint adoption of SLU practices and the associated benefits at the farm level. Furthermore, the joint adoption of SLU practices significantly influences downside production risk exposure (skewness), reducing the likelihood of crop failure or revenue loss. This highlights the potential of SLU practices to mitigate production risks arising from climate variability and extreme environmental events, such as droughts, high temperatures, and heatwaves. Our findings also provide a detailed breakdown of the adoption of SLU practices and their subsequent impacts across different agro-ecological zones. Specifically, farmers in the maize-wheat and rice-wheat zones experience greater economic benefits from the joint adoption of SLU practices compared to those in the cotton-mixed zone. Notably, the joint adoption of SLU practices significantly reduces downside production risk across all agro-ecological zones. Thus, these findings highlight the effectiveness of SLU practices in enhancing resilience and supporting farm income in the face of extreme climate variability. Nevertheless, this study contributes to the growing body of literature on adopting SLU practices and their subsequent impact on farm performance, particularly in developing countries where climate variability, limited access to information, and production risks threaten resilience and exacerbate rural poverty.

Our findings indicate that the adoption of SLU practices significantly enhances the ecological dimension by improving land quality, integrated pest management, and ecological farm management and reducing NPK fertilizer consumption per hectare. In the social dimension, joint adopters show significant improvements in income sources, sustainable knowledge management, and access to extension services. Further, regarding the economic dimension, our findings confirm that joint adoption of SLU practices significantly boosts farm mechanization, crop yields, farm income, and input sustainability. This encourages farmers to invest in land and farm improvements, leading to higher productivity, improved performance, and sustainable rural livelihoods with increased income.

This study outlines key policy actions to promote the broader adoption of SLU practices in Pakistan, with a special focus on integrating rural and remote farmers. First, policies aimed at improving agricultural productivity, increasing farm revenue, and reducing production risks should focus on overcoming barriers to SLU adoption. For instance, the federal government (e.g., the Ministry of National Food Security and Research) should collaborate with local seed companies to enhance farmers’ access to drought-tolerant crop varieties and affordable herbicides. Additionally, integrating climate adaptation and SLU practices into the broader national agenda, along with developing climate action plans for agriculture, is essential. Notably, targeted incentives, such as subsidies and financial support for climate-resilient crop varieties, should be offered to encourage SLU adoption. Second, scaling up public sector extension services is crucial for disseminating practical knowledge and technology to farmers using ICT. Community-based partnerships should be established to support climate action, with task forces working together to provide rural communities with essential information, including short-term weather forecasts, seasonal forecasts, early warning systems, and historical climate data. This approach would accelerate the adoption of SLU practices and help farmers select the most effective strategies for their specific regions. Since choosing the right combination of SLU practices requires reliable information and expert guidance, these services are vital for informed decision-making. Third, rural investments, including water harvesting systems and flood defenses, are necessary to reduce vulnerability. Additionally, capacity building through training programs and the establishment of local learning centers focused on SLU practices and climate-smart agriculture will foster an environment conducive to sustainable agricultural transformation. By implementing and integrating these measures, Pakistan can realize a shift towards climate-smart agriculture, particularly in climate-vulnerable areas where crop failure, declining revenues, and food insecurity increasingly threaten farm income and rural livelihoods.

This study makes a valuable contribution to addressing the stated objective, though it has some limitations due to the use of cross-sectional survey data. Despite these limitations, sour findings effectively rule out the presence of systematic bias. First, a panel dataset would be more suitable for capturing the long-term impacts of SLU practices and their effect on skewness, as certain SLU practices (e.g., minimum tillage and soil health improvement through organic matter) require several years to yield economic benefits. To enhance the robustness of the analysis, we recognize the need to expand the scope in future studies by including more diverse regions and populations. This would help validate the findings presented here and provide a broader perspective. Third, an additional proxy for farmers’ risk preferences could provide more precise estimates of production risk exposure, but due to data constraints, this was not feasible. Future research could incorporate such a proxy to examine farmers’ exposure to production risk further.