The concept of agricultural sustainability

From a purely anthropocentric perspective, which seeks to meet the food and fiber needs of the current generation without jeopardizing the capacity of future generations to do the same (WCED, 1987), sustainability in agriculture is a multidimensional concept defined by the three pillars of the Brundtland Commission’s sustainable development concept: environmental, economic, and social, each of which is essential for ensuring the long-term viability of farming systems. Environmentally, sustainable agriculture focuses on preserving natural resources, enhancing soil health, conserving biodiversity, and mitigating harmful practices such as overexploitation of water and greenhouse gas emissions (Campanhola and Pandey, Reference Campanhola and Pandey2018; Muhie, Reference Muhie2022). Economically, it ensures that agricultural activities remain profitable for farmers, resilient against market volatility and climate variability, and capable of sustaining livelihoods (Pretty, Reference Pretty2008). Socially, it emphasizes equitable access to resources, fair labor practices, and the well-being of rural communities (Timmermann and Félix, Reference Timmermann and Félix2015).

From a more holistic perspective, which sees sustainability not only as a theoretical concept about humans ‘living in harmony with nature and with one another’ (Mebratu, Reference Mebratu1998) over generations, but also ‘as an objective feature of the world, a numinous condition that makes life on planet Earth possible and meaningful for this and future human generations’, and ‘which implies that life in all forms is precious, it is worth sustaining’ (Sabau, Reference Sabau2024), sustainability in agriculture receives essentially a new connotation. By viewing sustainability as an intrinsic value inherent in the interconnected systems of life on Earth, those practicing agriculture will pay attention not only to what humans can extract from the Earth or what they can dispose of as waste in the terrestrial and ocean environments, but also to how they need to participate responsibly in the maintenance of the life-web at the core of sustainability. The farm’s sustainability will be measured not only by its economic efficiency but also by its capacity to exist and function as a social-ecological system in the long-term (Ostrom, Reference Ostrom2009, Reference Ostrom2014). This requires scientific knowledge of how ecosystems work, and how humans can benefit from protecting the structures, functions, and processes specifically embedded in the ecosystems that sustain life, by organizing farming activities to work with nature and not against it, and by observing the laws of nature (Georgescu-Roegen, Reference Georgescu-Roegen1971) and the planetary boundaries (Rockström, Steffen and Noone, Reference Rockström, Steffen and Noone2009; Steffen et al., Reference Steffen, Richardson, Rockström, Cornell, Fetzer, Bennett, Biggs, Carpenter, De Vries, De Wit, Folke, Gerten, Heinke, Mace, Persson, Ramanathan, Reyers and Sörlin2015). It also requires changes in human behavior, enabling farmers to promote environmental stewardship and social cooperation. Some of the ethical principles that farmers need to consider for a life of sustainability are: contentment, which calls for living harmoniously within ecological limits; justice, which broadens the scope of fairness to include ethical obligations not only to their families and communities, but also to all forms of life; and meaningful freedom, advocating for responsible actions within ecological and moral boundaries (Sabau, Reference Sabau2024). Similarly, Ekardt (Reference Ekardt2024) expands the sustainability discourse by emphasizing transformation, governance, ethics, and legal frameworks as critical dimensions that can be applied in practicing sustainable agriculture. Ekardt (Reference Ekardt2024) argues that achieving true sustainability requires systemic changes that extend beyond technological advancements. This includes rethinking human beings’ essential needs, and considering frugality as consumers, redesigning governance systems by embedding sustainability into legal and policy structures, and addressing ethical imperatives such as intergenerational justice and global equity (Ekardt, Reference Ekardt2024). Together, these perspectives deepen the understanding of agricultural sustainability as a comprehensive approach whose aim is not only to enhance productivity but also to ensure the protection of ecosystems, to foster social equity, and to support the ethical stewardship of resources. By integrating environmental, economic, social, ethical, and governance dimensions, sustainable agriculture becomes a transformative pathway toward securing long-lasting agrifood systems and livelihoods for farmers while safeguarding the planet’s ecological balance for future generations.

While sustainability has deep ethical and philosophical roots, as highlighted by Sabau (Reference Sabau2024) and Ekardt (Reference Ekardt2024), this study operationalizes it using empirical farm-level efficiency measures. We define agricultural sustainability through the three pillars: economic, environmental, and social. These are quantified through a DEA-based framework to assess how farms manage resources while aligning with ecological and economic objectives. This pragmatic lens helps translate abstract sustainability concepts into actionable indicators, which are essential for policy decision-making.

The importance of achieving efficiency in agriculture

Technical efficiency refers to the ability of farms to achieve the maximum possible output given the resources available. It measures operational performance and identifies whether farms are fully utilizing their inputs, such as labor, land, and machinery, sunshine, rainfall, and traditional knowledge. Allocative efficiency, on the other hand, evaluates whether the mix of inputs is being utilized in a cost-effective manner, considering their relative prices and marginal productivity (Farrell, Reference Farrell1957). Cost efficiency is an amalgamation of technical and allocative efficiencies, reflecting the overall economic performance of a farm. Scale efficiency examines whether farms are operating at the optimal size to maximize productivity and minimize costs, offering insights into the potential benefits of adjusting operational scale (Charnes, Cooper and Rhodes, Reference Charnes, Cooper and Rhodes1978). Environmental efficiency considers not only inputs and outputs but also the environmental impacts of farming processes (Fraanje et al., Reference Fraanje, Garnett, Röös and Little2019). These efficiency metrics play a crucial role in advancing agricultural sustainability when farmers make deliberate choices to optimize resource use, enhance economic viability, and protect ecological resilience on the farm (De Koeijer et al., Reference De Koeijer, Wossink, Smit, Janssens, Renkema and Struik2003).

Achieving these efficiencies is critical for agricultural sustainability, especially in regions like Western Newfoundland, where environmental and socio-economic challenges require innovative farming strategies. Improvements in efficiency can enhance profitability for farmers, reduce resource wastage, and mitigate environmental impacts, thereby supporting the long-term sustainability of farming systems. Understanding these efficiencies is particularly relevant for informing evidence-based policies and promoting resilient agricultural practices in the face of climate change and resource constraints (Färe et al., Reference Färe, Grosskopf, Norris and Zhang1994). Table 1 provides an overview of the efficiency types used in the study.

Table 1.

Summary of efficiency types used in the study

Recent studies have increasingly employed the DEA tool to assess sustainability from a multidimensional perspective. For example, Ait (Reference Ait2023) proposed a DEA-based framework to jointly evaluate economic, environmental, and social performance at the farm level, offering a comprehensive sustainability efficiency metric. Kyrgiakos et al. (Reference Kyrgiakos, Kleftodimos, Vlontzos and Pardalos2023) conducted a systematic review of DEA applications in agriculture and highlighted emerging trends and methodological innovations under the lens of sustainability. Similarly, Dong, Mitchell and Colquhoun (Reference Dong, Mitchell and Colquhoun2016) combined principal component analysis with DEA to assess improvements in soybean production systems in the U.S. Midwest, emphasizing resource optimization and environmental stewardship. These studies underscore the versatility of the DEA method in sustainability analysis and validate its growing role in agri-environmental assessments.

The agricultural context of Western Newfoundland

The study area, Western Newfoundland, lies in the NL province of Canada. It is an island, part of the boreal ecozone, characterized by acidic soils, rugged terrain, and a harsh climate. The region’s boreal climate features cool summers, cold winters, and a short growing season of approximately 3–4 months, as well as high annual precipitation, including significant snowfall in winter, which affects soil moisture recharge and drainage (AAFC, 2021). These climatic conditions, combined with frequent freeze–thaw cycles, create challenges for soil health, crop growth, and farm management. As a result, the province experiences the challenges of the highest food insecurity among all Canadian provinces, with 23% of the province’s families being food insecure in 2022 (Statistics Canada, 2023). The province has about 300 different farms, and most of the farms in the province are small-scale, with an average farm size of 144 acres, which produce a limited number of commodities, including dairy products, chicken, eggs, greenhouse and nursery produce, vegetables, and berries (NL, 2019). In 2019, food self-sufficiency in the province was assessed to be 10–12%, and at that time the provincial government set the target of doubling food self-sufficiency in the province by 2022 (NL, 2018).

The sustainability challenges faced by Newfoundland share similarities with other global agricultural systems facing climatic and resource-based constraints. For example, Gong et al. (Reference Gong, Yin, Chen, Zhang, Tian, Wang, Wang and Cui2025) demonstrated that optimizing soil phosphorus status can significantly reduce fertilizer use, which aligns with the findings of this study that highlight chemical fertilizer overuse as a major inefficiency. Similarly, Wu et al. (Reference Wu, Liu, Cheng, Gu, Guo and Jiao2025) showed how improved evapotranspiration modeling can inform better irrigation practices, which is relevant to boreal regions where water management is seasonally critical. Moreover, international efforts to enhance agroecosystem resilience in saline and degraded landscapes (Hu et al., Reference Hu, Zhao, Hu, Qi, Suo, Pan, Song and Chen2022; Du et al., Reference Du2024) parallel local needs for soil restoration and climate-adapted farming in Western Newfoundland. These global advances underscore the importance of localized efficiency strategies informed by broader scientific progress in agro-environmental sustainability.

Beyond agro-climatic constraints, Newfoundland presents unique challenges related to food security, rural economic development, and geographic isolation. With one of the lowest food self-sufficiency rates in Canada (10–12%) and 23% of households experiencing food insecurity in 2022 (Statistics Canada, 2023), there is a growing need to enhance local agricultural productivity and resilience. This context makes Western Newfoundland an important case study for exploring regionally tailored strategies to support sustainable agriculture through efficiency improvements.

Study area

The study area is Western Newfoundland (Figure 1), the island portion of the NL province in Atlantic Canada. The region is characterized by podzolic soils (low fertility) and a cool summer boreal climate that significantly influences its agricultural production. The predominant soils are acidic podzols, often low in nutrients and prone to leaching, requiring amendments for optimal crop growth. Poorly drained gleysols are also common in the area’s valleys and wetlands, affecting soil water retention. The region’s boreal climate is characterized by arctic summers, freezing winters, and high annual precipitation ranging from 1100 to 1400 mm, with significant snowfall in winter. A short growing season of 90–120 days makes timing critical for planting and harvesting. Vegetables and root crops thrive in the region due to their adaptability to acidic soils and cooler temperatures, though careful management of soil pH, drainage, and nutrient levels is essential. There are 37 farms in the research area, each producing a combination of field and greenhouse vegetables, herbs, berries, fruit, flowers, hay, and nursery sod.

Figure 1.

This study area illustrates the geographical location of Western Newfoundland, NL, Canada, highlighting key agricultural areas and the distribution of farms within the region.

Farm efficiency as a measure of sustainability

Assessing sustainability in agriculture is inherently complex, given the many ways in which crops and livestock sustain human populations and the diverse pathways by which farming can be managed to balance profitability with environmental stewardship (Färe et al., Reference Färe, Grosskopf, Norris and Zhang1994; Harris and Fuller, Reference Harris, Fuller and Smith2014). In this study, we adopt the premise that farm-level productive efficiency is a useful proxy for sustainability, since efficiency gains often result from reducing the use of inputs such as chemical fertilizers and pesticides that typically generate negative environmental impacts.

The methodological foundation of this research builds on the seminal work of Farrell (Reference Farrell1957), later advanced by Afriat (Reference Afriat1972) and Charnes, Cooper and Rhodes (Reference Charnes, Cooper and Rhodes1978), and subsequently refined by Färe et al. (Reference Färe, Grosskopf, Norris and Zhang1994). These developments established DEA as a robust non-parametric approach for measuring efficiency. While the technical underpinnings of DEA are well established (Coelli, Rao and Battese, Reference Coelli, Rao and Battese1998), the novelty of this study lies in applying a multidimensional DEA framework that integrates both economic and environmental dimensions of sustainability in a boreal agricultural context.

This research evaluates five efficiency measures, each representing a distinct sustainability dimension:

• • Technical efficiency (TE): defines the ability of farms to convert inputs (land, labor, seed, fertilizer, fuel) into outputs (revenue, yield). A score of 1 indicates full efficiency, while values below 1 reflect inefficiencies (Farrell, Reference Farrell1957).

• • Allocative efficiency (AE): assesses whether farms select the most cost-effective mix of inputs given their relative prices (Farrell, Reference Farrell1957; Coelli et al., Reference Coelli, Rao, O’Donnell and Battese2005).

• • Cost efficiency (CE): combines TE and AE to capture farms’ overall economic performance, reflecting whether they minimize costs while maintaining outputs.

• • Scale Efficiency (SE): examines whether farms operate at an optimal size, distinguishing between inefficiencies due to under-scaling (IRS) or over-scaling (DRS) (Charnes, Cooper and Rhodes, Reference Charnes, Cooper and Rhodes1978).

• • Environmental efficiency (EE): extends DEA to include ecological impacts of inputs (e.g., CO₂ emissions from fertilizer and fuel use), assessing whether farms minimize environmental burdens per unit of output (De Koeijer et al., Reference De Koeijer, Wossink, Smit, Janssens, Renkema and Struik2003; Ait, Reference Ait2023). To calculate EE, we derived the environmental burden of each input from standardized impact coefficients found in peer-reviewed life-cycle assessment (LCA) studies and public databases. For example, carbon footprint values for nitrogen fertilizer were estimated at 5.88 kg CO₂e per kg of active ingredient (Gallardo, Reference Gallardo2024), while pesticide toxicity was weighted using a composite toxicity index (De Koeijer et al., Reference De Koeijer, Wossink, Smit, Janssens, Renkema and Struik2003). Diesel fuel consumption was converted to CO₂e using 2.68 kg CO₂e per liter. Where data were reported per unit area (e.g., water use per hectare), values were normalized based on reported farm sizes. These environmental impact equivalents were used as input proxies in the DEA model, replacing raw input quantities to estimate EE scores that reflect ecological performance. This approach aligns with the environmental DEA methods used in recent sustainability studies (Ait, Reference Ait2023; Kyrgiakos et al., Reference Kyrgiakos, Kleftodimos, Vlontzos and Pardalos2023).

Detailed mathematical formulations, linear programming models, and illustrative figures supporting these efficiency measures are provided in Supplementary Material. By relocating technical details outside the main text, this study is more easily readable. Besides, it highlights its novel contribution: demonstrating how a multidimensional DEA framework can integrate environmental sustainability with farm-level efficiency analysis, providing context-specific insights for agricultural policy and practice in Newfoundland.

Data collection

The agricultural study conducted in Western Newfoundland involved contacting all 37 crop farms in the region during September–November 2024. Fifteen farms agreed to participate, yielding a response rate of approximately 40%. Data collection was achieved through structured questionnaire surveys, focusing on both input and output metrics. Input data encompassed resource usage, including land area, labor, water, fertilizer and pesticide usage, and energy consumption. On the output side, data were gathered on farm productivity and sustainability indicators such as crop yield, economic returns, and environmental indicators, like soil health and carbon emissions, and social indicators, including labor conditions and community impact. The survey instrument included 32 closed-ended questions, divided into five thematic sections: general farm characteristics, production inputs, environmental practices, economic performance, and labor dynamics. The questionnaire was pilot tested with two local farmers and one agricultural extension officer to ensure clarity, relevance, and data quality. Based on their feedback, minor revisions were made to improve survey usability. This structured approach ensured the collected data were both consistent and representative of the farms’ operational practices, providing an overview of agricultural sustainability in the region.

Participation was voluntary, though participants were promised access to a summary of results and personalized efficiency profiles. Missing data were minimal and were addressed using pairwise deletion in non-critical fields; however, farms with missing data in key input or output variables required for DEA modeling were excluded from specific model runs. Consistency screening and normalization of the final dataset were conducted using the R programming language.

Decision-making units

The decision-making units (DMUs) in this study are the 15 individual farms surveyed in Western Newfoundland. Each farm as one DMU operates independently, employing various agricultural practices and resource management strategies. The analysis evaluates the efficiency of these farms by comparing their input usage against output production, providing insights into the effectiveness and sustainability of their operations. Although the 15 participating farms produced a diverse mix of crops, they shared key characteristics that ensured the methodological homogeneity required for DEA. All farms grow a combination of field and greenhouse vegetables (e.g., potatoes, carrots, lettuce), berries, and herbs, and follow similar production goals focused on yield and revenue optimization. Thus, despite crop variety, the output structure (measured in revenue and residue) and resource use patterns were sufficiently similar to justify their inclusion as comparable DMUs under the DEA method.

Implementation of DEA

An input-oriented DEA model was selected for this study because the primary goal was to evaluate how farms could minimize input use while maintaining their current output levels. In the context of Western Newfoundland, where resources such as labor, fertilizer, and energy are often limited, expensive, or environmentally sensitive, optimizing input usage is critical for improving sustainability. Moreover, output levels (e.g., crop yields or revenue) are often constrained by environmental factors like climate and soil conditions, which farmers have limited control over. Therefore, an input minimization approach aligns more realistically with farmers’ operational decision-making and offers practical guidance for enhancing efficiency without requiring significant changes in output targets.

The selection of input and output variables was guided by their relevance to evaluating agricultural sustainability from an economic and environmental perspective. Inputs such as land, labor, machinery, fertilizer, pesticide, seed, fuel, and electricity represent major resource expenditures that impact both cost-efficiency and environmental performance. These inputs are commonly used in DEA studies of farm efficiency (Ait, Reference Ait2023; Grzelak and Kryszak, Reference Grzelak and Kryszak2023) and align with sustainability goals by reflecting areas where resource optimization can reduce economic costs and ecological impacts.

Outputs included revenue (economic output), yield (agronomic productivity), and crop residue (a proxy for biomass and soil health contribution). These indicators were chosen to capture the dual goals of productivity and ecological stewardship. The residue, in particular, provides insight into soil carbon input and organic matter recycling, which are both critical to long-term environmental sustainability. This variable mix supports a holistic assessment of the farms’ ability to balance output maximization with responsible input use.

To evaluate agricultural efficiency and sustainability in Western Newfoundland, a structured DEA was implemented. Both constant returns to scale (CRS) and variable returns to scale (VRS) DEA models were applied to assess scale efficiency and better capture the diversity in farm sizes. While no formal statistical test (e.g., likelihood-ratio test or bootstrapping) was performed due to sample size constraints, the model outputs were compared across CRS, VRS, and non-increasing returns to scale (NIRS) specifications to evaluate consistency and identify the nature of scale inefficiencies. The use of both CRS and VRS models is consistent with recommendations for small farm sample analysis in agricultural DEA studies (Coelli et al., Reference Coelli, Rao, O’Donnell and Battese2005).

Input and output data were normalized to maintain consistency across datasets, and efficiency scores for each DMU were calculated using the ‘Benchmark’ and ‘deaR’ packages in the RStudio Integrated Development Environment (IDE) (Coll-Serrano, Vicente and Benito, Reference Coll-Serrano, Vicente and Benito2023; Bogetoft and Otto, Reference Bogetoft and Otto2024). This package streamlined the process by enabling seamless data integration, efficient computation of DEA scores, and application of both VRS and CRS models. It also provided visualization tools to depict efficient frontiers and highlight best-performing farms, allowing for an intuitive understanding of the results. The analysis identified areas of inefficiency and generated actionable insights to enhance farm productivity and sustainability. By incorporating both CRS and VRS approaches, the study offered a holistic understanding of technical and scale efficiencies. This dual approach emphasized resource use optimization and supported the development of sustainable agricultural practices tailored to the specific needs of the region. Additionally, a stepwise regression model was employed to identify key factors influencing farm efficiency (Żogała-Siudem and Jaroszewicz, Reference Żogała-Siudem and Jaroszewicz2021). Stepwise regression is a systematic statistical method that iteratively selects or removes predictor variables based on predefined criteria, such as the Akaike information criterion (AIC) or p values, to improve model performance and reduce multicollinearity (Smith, Reference Smith2018). In this study, stepwise regression was used to determine the most significant socioeconomic and environmental variables affecting technical, allocative, cost, and environmental efficiencies.

Regarding sample size, while DEA does not require large samples, using only 15 DMUs can increase sensitivity to outliers and reduce discriminatory power. This limitation was mitigated by (i) selecting a parsimonious set of inputs and outputs; (ii) ensuring homogeneity among DMUs; and (iii) complementing DEA scores with regression analysis to strengthen interpretability. Nevertheless, the relatively small sample size is acknowledged as a constraint and an area for future study expansion.

This approach enabled a more targeted analysis of efficiency determinants, allowing for the identification of critical factors such as farmer experience, farm size, education level, and environmental practices that impact overall farm performance. The combination of DEA and stepwise regression provided a robust framework for evaluating agricultural efficiency while offering actionable insights to optimize resource use and enhance sustainability.