Dairy cattle production systems are increasingly scrutinized for their sustainability amid a broader reassessment of intensive production models. These concerns encompass environmental, economic and social dimensions. Environmentally, they relate to greenhouse gas emissions and the inefficient use of natural resources. Economically, sustainability involves farm profitability, resilience to market fluctuations and dependence on policy support, within a context where viability is also closely linked to social acceptability. Socially, it includes animal welfare, farmers’ quality of life and the vitality of rural communities (Broom, 2023; Pavanello et al., 2024). Within this framework, livestock production systems in desert regions are particularly debated due to their structural constraints and implications for sustainable development in these especially fragile environments (Alary et al., 2020).
       
In Algeria, agricultural development policies have promoted the intensification of dairy cattle production and its expansion into Saharan regions. This process has led to the emergence of new livestock production models; however, these systems remain constrained by significant environmental and socio-economic limitations in these formerly marginal areas, which are characterized by extreme climatic conditions. This situation raises concerns about their sustainability (Senoussi et al., 2010; Eulmi et al., 2024). Nevertheless, the actual performance of these systems has not been sufficiently assessed. In this context, this study aims to analyze the sustainability and develop a typology of dairy cattle farming systems in two Algerian Saharan regions (Biskra and Ghardaïa) to better understand their operational logic, evaluate their efficiency and protect the environment and natural resources for future generations.
 

This study was conducted in two arid zones in the northern Algerian Sahara (Fig 1): the regions of Biskra (34.85°N, 5.72°E; 124 m altitude) and Ghardaïa (32.49°N, 3.67°E; 500 m altitude). These regions are characterized by contrasting climatic conditions, with average temperatures of 23.1°C and 22.3°C and annual rainfall levels of 152 mm and 95 mm, respectively. They were selected due to the lack of previous sustainability studies in these areas and their contrasting agricultural dynamics, characterized by intensive crop-oriented systems in Biskra and more livestock-oriented production systems in Ghardaïa.


       
The farms surveyed were randomly selected from official lists provided by the Agricultural Services Directorates of Biskra and Ghardaïa. The selection was based on inclusion criteria, including participation in dairy development programs and possession of the required veterinary health certificate (≥6 dairy cows). The final sample comprised 92 farms, including 32 in Biskra and 60 in Ghardaïa. These farms accounted for 71.1% and 60%, respectively, of all dairy cattle farms recorded in 2023-2024 in the two study regions (excluding the selection criteria applied in this study).
       
The survey was conducted face-to-face between October 2023 and April 2024. A questionnaire consisting of 180 items was used to assess farm sustainability using the IDEA3 method, chosen for its accessibility, scientific relevance and relative simplicity of implementation compared to alternative approaches.
       
The IDEA3 method (2008 version) is a French tool for assessing agricultural sustainability based on three dimensions: agroecological, socioterritorial and economic. It is structured around 42 indicators (Fig 2) grouped into 10 components (Table 2), scored using a predefined grid with specific thresholds. The overall sustainability level of a farm corresponds to the lowest score among the three dimensions and a farm is considered sustainable when its final score exceeds 50% of the theoretical maximum (Vilain et al., 2008).

​
       
To standardize the results and facilitate interpretation, the sustainability scores of each component were normalized and expressed as a percentage of the theoretical maximum score (TMS) using the following formula:
 

  

       
The economic dimension of the IDEA3 framework was not assessed due to the lack of reliable economic data and the confidential nature of such information within the studied systems. To address this limitation, farmers were surveyed about the primary economic constraints likely to impact the long-term sustainability of their farms.
         
Statistical data processing and graphical presentation of the results were carried out using RStudio (version 4.4.2). The analyses included namely one-way analysis of variance (ANOVA), Welch’s ANOVA, and the Kruskal–Wallis test. Post hoc multiple comparisons were performed usingTukey’s honestly significant difference (HSD) test or Dunn’s test with Bonferroni correction, as appropriate. Statistical significance was set at p < 0.05.Given the high dispersion and non-normal distribution of several variables, farm structure and characteristics were described using medians and interquartile ranges (Q1-Q3), which more appropriately represent the distributions of these variables than arithmetic means.
       
A structural typology of livestock farms was developed using eight active quantitative variables: labor (Lb), utilized agricultural area (UAA), cultivated agricultural area (CAA), arable crop area (ASA), permanent crop area (PPA), available forage area for livestock (AFL), cattle herd (CA), dairy cow (DC) and stocking rate (SR). The data was first corrected for normality using the Yeo-Johnson test before undergoing principal component analysis (PCA) with Varimax rotation. This was followed by hierarchical ascending classification (HAC) using Ward’s method. A second typology was developed to classify farms based on their sustainability. This typology utilized six active quantitative variables that corresponded to the six components of the agro-ecological and socio-territorial dimensions. 
       
The suitability of the data for principal component analysis (PCA) was evaluated using the kaiser-meyer-olkin (KMO) index and the variance explained by the factorial axes. Variable selection was based on agronomic relevance, the KMO index, results from the anti-image matrix and each variable’s contribution to the factorial axes. The typology was then established by determining the dendrogram cut-off level, taking into account both the statistical consistency and the agronomic interpretability of the resulting groups.

Classification of the surveyed cattle farms
 
Principal component analysis identified nine factorial axes, of which the first two explained 83% of the total variance. The first axis (59.6%) was mainly associated with land resources and cropping orientation, while being negatively correlated with labor and stocking rate, reflecting a gradient between land-extensive and labor-intensive production systems (Fig 3). The second axis (23.4%) was primarily related to cattle and dairy herd size, highlighting differences in livestock specialization among farms.


       
Based on the distribution of farms along the first two principal components, hierarchical clustering identified five distinct farm typologies (Fig 4).


       
The typological analysis revealed significant structural heterogeneity among dairy cattle farming systems (Table 1). Three primary gradients emerged: farm size, livestock intensification and labor organization. A land-size gradient contrasted small farms (C1 and C5) with large farms (C3 and C4), while C2 represented an intermediate profile. Livestock intensity also varied markedly among classes, being highest in C5 and lower in C3 and C4 despite their larger herds. Labor organization ranged from predominantly family-based systems in C1 to more externalized structures relying on hired labor in C4. A clear territorial differentiation was also observed, with C3 mainly associated with Biskra, whereas C2 and the exclusively Ghardaïa-based C5 were more representative of the latter region.


 
Sustainability levels of the surveyed cattle farms
 
Principal component analysis (PCA) revealed six axes explaining 100% of the total variability in the sample. Axis 1, which accounts for 32.9% of the total inertia, is positively associated with components of the socioterritorial dimension (Fig 5). Axis 2, which accounts for 24.2% of the total inertia, reflects the level of agroecological sustainability. There is a positive correlation for the first two components of this dimension-domestic diversity and spatial organization-and a negative correlation for the third component, agricultural practices.


       
The classification of farms based on the first two principal component analysis axes and hierarchical clustering revealed six distinct classes (Fig 6).


       
The typological analysis identified six farm classes grouped into three sustainability profiles (Table 2). The first profile (Classes A and E) comprised low-sustainability systems, with scores below the 50% threshold in most dimensions, indicating significant structural vulnerabilities. The second profile (Classes B, C and D) exhibited intermediate and varied sustainability, combining relatively satisfactory agroecological performance with persistent socio-territorial limitations. Within this group, Class C represented a transitional profile, with values approaching the sustainability threshold. The third profile, represented by Class F, demonstrated the highest level of sustainability, characterized by scores generally exceeding the 50% reference threshold and a better balance between agroecological and socio-territorial dimensions.


 
Sustainability levels of the structural classes of the surveyed cattle farms
 
The sustainability assessment across structural classes reveals significant heterogeneity (Table 3). Agroecological performance (D-AE) remains relatively consistent across classes, with values near the 50% threshold. In contrast, socioterritorial performance (D-ST) exhibits greater variation, primarily due to low U4 and U6 scores in most classes. Class C4 distinctly stands out with the highest socioterritorial and overall sustainability levels, indicating a more advanced transitional profile. Additionally, classes C1 and C3 display the lowest sustainability levels, while C2 and C5 show intermediate profiles characterized by moderate agroecological performance but persistent socioterritorial constraints.


 
Economies impacts
 
The assessment of the economic constraints faced by farms in the Saharan regions under study identified nine types of difficulties, whose reporting frequencies varied according to their nature (Fig 7). The high cost of animal feed was the most frequently reported constraint, with a mean reporting frequency of 90.2 ± 16.5% across the different groups. It was followed by the high cost of cattle (52.1 ± 21.1%) and difficulties in accessing subsidies (51.4 ± 21.1%). Other constraints included the costs of agricultural services and inputs (25.2 ± 15.5%), electricity (19.9 ± 13.5%), livestock equipment (8.2 ± 7.0%), and labour (5.0 ± 6.7%). These difficulties were compounded by economic instability resulting from anthropogenic or natural events (6.9 ± 5.4%) and limited access to dairy cow insurance (4.4 ± 7.8%). The comparative analysis of the typological profiles also revealed marked heterogeneity in the cumulative reporting frequencies of economic constraints. Groups C4 and F exhibited the lowest cumulative reporting frequencies, with four and six of the nine identified constraints reported, respectively, corresponding to cumulative frequencies of 166.7% and 183.3%. Conversely, class D recorded the highest cumulative reporting frequency (391.7%), with all nine identified constraints reported, followed by group C5, which reported eight out of the nine constraints, with a cumulative frequency of 381.3%.


 
       
The typology reveals structurally comparable farm profiles across regions, with differences primarily driven by their spatial distribution. This structural organization aligns with regional agricultural orientations reported in the literature. In Biskra, class C3 represents a predominantly crop-oriented system (Naouri et al., 2015), whereas in Ghardaïa, class C2 corresponds to an intermediate mixed system and class C5-exclusively observed in this region-indicates a stronger focus on cattle production (Senoussi et al., 2010). These patterns underscore the combined influence of agroecological conditions and socioeconomic factors on farm specialization processes (Mokhnane et al., 2026).
       
The sustainability typology highlights that different farm profiles generally share similar strengths and weaknesses. At the agroecological level, most farms are characterized by high levels of animal diversity (A3) and genetic heritage valorization (A4), two key factors typically associated with greater resilience to economic and environmental shocks and reduced production risk (Alary et al., 2020; Pavanello et al., 2024). This trend is further reinforced by strong valorization of organic matter flows (A7 and A13), which supports resource recycling and soil fertility maintenance (Broom, 2023; Pavanello et al., 2024; Wongkrachang and Anurut, 2025).
       
However, these strengths are offset by limited autonomy in managing productive resources and a strong dependence on external inputs. The poor performance observed in land management and forage area utilization (A5, A10 and A11) indicates an incomplete valorization of local resources, which may restrict the feed self-sufficiency of farms. This issue is particularly critical in the arid zones under study, where climatic constraints exacerbate the vulnerability of livestock systems (Benmihoub et al., 2021). Additionally, reliance on external production inputs (A12 and A15), combined with limited consideration of ecological functions and resources (A8, A9, A16, A17 and A18), reflects a predominantly production-oriented strategy that may increase environmental pressures and jeopardize long-term sustainability (Alary et al., 2020; Unnikrishnan et al., 2022; Broom, 2023; Pavanello et al., 2024).
       
A similar pattern is observed in the socioterritorial dimension. The farms studied generally demonstrate significant contributions to employment (B9), primarily driven by the use of family labor-a characteristic widely reported in livestock systems (MADR, 2012; Zenchi and Abdoun, 2015; Sowmya et al., 2022). They also exhibit relatively high scores for activity continuity (B11), reflecting the family-based and intergenerational nature of cattle farming in Algeria. Additionally, relatively high levels of quality of life and low social isolation within the community (B16 and B17) underscore the strong and enduring attachment of farmers to this activity, extending beyond its purely economic function (Madelrieux et al., 2010; MADR, 2012; Zenchi and Abdoun, 2015).
       
Despite these strengths, socioterritorial sustainability remains constrained by several persistent limitations. The near absence of quality-oriented approaches (B1) reflects weak product differentiation in a context where concerns about food availability continue to take precedence over certification or labeling strategies (Senoussi et al., 2010). Low scores in non-organic waste management (B3) also indicate limited consideration of certain environmental externalities associated with livestock production. Similarly, the weak contribution to the global food balance (B12) reflects the predominantly local orientation of the systems under study (Vilain et al., 2008). Furthermore, knowledge sharing and access to training (B14) remain limited, which may hinder the diffusion of innovations and improved practices; this constraint has already been reported in Algerian cattle systems (Bekhouche-Guendouz,  2011). Finally, low performance in work organization (B15), observed in all classes except for class F, reflects persistent labor overload, which may limit adaptive capacity and farm improvement (Madelrieux et al., 2010; Ouakli, 2018).
       
The typological analysis identified six distinct farm profiles, of which only Class F achieves a clearly sustainable status. Within this framework, Class C emerges as the profile closest to sustainability, demonstrating intermediate yet relatively balanced performance across dimensions. Although it falls below the sustainability threshold, its levels can be considered moderately satisfactory, indicating a transitional profile with potential for improvement through targeted management adjustments.
       
When compared with studies conducted in other Algerian regions, the highest sustainability class observed in this study (Class F) exhibits a higher score than those reported for the best sustainability classes of dairy cattle systems in the Cheliff Valley (37.0±5.5%; Ouakli et al., 2018), the Mitidja and Annaba regions (38.4±2.6%; Bekhouche-Guendouz,  2011) and the Algiers region (45.0±6.3%; Ikhlef-Mehennaoui,  2016). However, it remains slightly lower than the most performant class reported in cattle systems from the Sétif highlands (56.2±2.9%; Bir et al., 2011). These comparisons suggest that, while structural and management differences explain part of the variability in sustainability levels, broader territorial and environmental constraints play a decisive role in shaping dairy system performance. This further supports the view that sustainability should not be considered an absolute value but rather a context-dependent outcome shaped by interactions among biophysical, socio-economic and territorial factors, rather than as an intrinsic property of farming systems.
       
The socio-territorial dimension emerged as the primary limiting factor for 90.2% of the surveyed farms across all sustainability classes, corroborating findings from previous studies on dairy cattle systems in Algeria and Tunisia (Yakhlef et al., 2008; Bir et al., 2011; Bekhouche-Guendouz, 2011; Ikhlef-Mehennaoui, 2016; Ouakli et al., 2018; M’Hamdi  et al., 2017; Ounalli et al., 2025). This recurring pattern reflects structural characteristics of North African livestock systems, including limited collective organization, weak product differentiation, restricted knowledge exchange and low territorial integration. However, it may also partly result from the limited suitability of certain IDEA3 socio-territorial indicators, originally developed for European agricultural contexts, when applied to local farming realities. Consequently, the relatively low sustainability levels should be interpreted with caution, as they likely reflect both genuine socio-territorial constraints and a partial mismatch between the assessment framework and local conditions. Moreover, given the absence of a comprehensive evaluation of the economic dimension, the identified sustainability profiles should be regarded as partial rather than complete representations of farm sustainability.
       
When comparing sustainability levels with the structural profiles of the surveyed farms, very low results were observed, with statistically significant differences between classes. These outcomes are also influenced by the socio-territorial dimension, indicating the absence of a clearly superior and sustainable farming model based solely on the structural characteristics of these production units. These findings are consistent with those of Bekhouche-Guendouz (2011), who studied dairy cattle farms in the Mitidja and Annaba regions.
       
Regarding the economic dimension, the observed patterns align with the sustainability typology, the Class F  and Class C4 demonstrated greater economic stability and resilience compared to the other profiles. Production efficiency emerged as the primary constraint across all classes, highlighting the strong dependence of dairy farms on external inputs, particularly feed and imported cattle,within a context of volatile agricultural markets (Lazereg et al., 2020). Financial autonomy was identified as a second major limitation, linked to reliance on public support schemes and frequent payment delays. Together, these constraints reveal a systemic economic vulnerability driven by input dependency, market instability and limited financial autonomy, which collectively undermine the long-term resilience of dairy production systems.

This study contributes to the literature by proposing an integrated typological analysis of the sustainability of dairy farming systems in an arid context, thereby enhancing the understanding of profile differentiation and structural constraints within an environment that remains insufficiently documented. The analysis reveals, on one hand, significant structural heterogeneity among farms and, on the other, the existence of six sustainability profiles. Among these, only Class F achieves a clearly sustainable level, while Class C exhibits an intermediate profile that is relatively promising but still insufficient.
       
These results demonstrate that sustainability arises from the interaction among agro-ecological, socio-territorial and economic dimensions. The agro-ecological dimension is characterized by dependence on inputs; the socio-territorial dimension by organizational and territorial weaknesses; and the economic dimension by vulnerability related to market instability and low financial autonomy.
       
In terms of policy recommendations, differentiated public policies should be implemented based on the levels of constraints, prioritizing the most vulnerable farms. These efforts should be complemented by awareness-raising initiatives aimed at promoting the adoption of sustainable and resilient practices.
       
From a methodological perspective, the IDEA3 approach, combined with typological analysis, effectively structures the diversity of systems; however, the absence of a comprehensive assessment of the economic dimension necessitates cautious interpretation and future adjustments to better reflect local realities.

The authors declare that they have no conflict of interest.