Spatial Typology of Lorena Avocado Production Systems in Colombian Lowlands (Casanare): Integrating Agronomic and Socioeconomic Characteristics

Abstract

Understanding the diversity of avocado production systems is crucial for developing effective agricultural policies and extension strategies. This study examined the Colombian avocado variety “Lorena” in the Colombian lowlands of Casanare through spatial typology analysis to inform sustainable agricultural development strategies. We employed spatial autoregressive modeling and clustering techniques to analyze data from 45 production systems, revealing heterogeneity despite small-scale operations with productivity (2.9 ton ha⁻¹) below regional (8 ton ha⁻¹) and national averages (11.03 ton ha⁻¹). Five distinct typologies emerged: transitional traditional (n = 15), intensive technical management (n = 4), experience-based traditional (n = 5), balanced management (n = 10), and comprehensive technical systems (n = 11). In contrast to conventional assumptions about economies of scale, productivity was not primarily determined by farm size, as smaller intensive technical management systems achieved the highest yields (3375 kg) despite having the smallest size (162.50 trees), followed by experience-based traditional systems (3280 kg). The spatial autoregressive model effectively captured spatial dependence in yield patterns, demonstrating the importance of geographic context in agricultural system analysis. Technology/practice adoption patterns varied markedly, with high adoption of established practices (>90%) but low foliar analysis adoption (17.78%). High organic fertilization adoption (93.33%) reflected a commitment to environmental sustainability but may partially explain productivity gaps, highlighting trade-offs between sustainability and short-term yield optimization. Socioeconomic analysis revealed characteristics of part-time farming systems, with 91.11% of producers having additional income sources and 95.56% using hired labor, suggesting evolved livelihood strategies that may enhance resilience. These findings challenge one-size-fits-all development approaches and demonstrate the need for tailored, spatially targeted interventions that account for specific production system characteristics, multiple pathways to sustainable intensification, and the complex interactions between productivity, sustainability, and socioeconomic factors in smallholder agriculture.

1. Introduction

Understanding the diversity of agricultural production systems through characterization and typification studies is crucial for sustainable agricultural development and achieving food security goals while maintaining environmental integrity [1,2]. In the context of global challenges, including climate change, biodiversity loss, and the need to feed a growing population, agricultural systems must transition toward more sustainable practices that balance productivity with environmental stewardship and social equity [3,4]. Sustainable agriculture integrates biological and ecological processes into food production, minimizes harmful non-renewable inputs, and utilizes farmers’ knowledge while making optimal use of environmental goods and services without compromising these natural assets [5]. This approach incorporates principles like integrated management, dynamic balance, regenerative design, and social development to create resilient agricultural systems that balance human nutritional needs with environmental quality and economic viability [6,7].

Farmers optimize productivity by making better decisions based on the complex interplay between non-controllable factors, such as climate and biological processes, and controllable crop management factors, while implementing proper planning and control tools [8,9]. This optimization involves understanding and managing the intricate interactions between natural resources, biological elements, technology, and the farmer’s managerial ability within the farm’s boundaries [10,11]. While these farm-level factors are important, agricultural productivity and sustainability are equally influenced by external elements beyond the farm’s boundaries that shape the broader operating environment. These external factors include market conditions that determine the demand and value of agricultural products; institutional elements such as agricultural policies, extension services, credit access, and insurance availability; and physical infrastructure including roads, storage facilities, and processing plants [12,13].

Colombia’s avocado sector exemplifies the potential for sustainable agricultural diversification in developing countries, with the country currently being the second largest global avocado producer with nearly 1 million tons annual output attributed to agroclimatic advantages, strategic geographic location, and land availability [14,15,16]. Although the “Hass” variety dominates the export market, local varieties like “Lorena” remain significant for domestic consumption and small-scale producers, representing an important but understudied segment of Colombia’s avocado sector [17].

Recent studies have highlighted the importance of adopting improved agronomic practices and technologies to achieve sustainable avocado production outcomes, balancing yield optimization with environmental considerations. Farms implementing sustainable technological approaches exhibit better yield performance, lower production costs, and reduced disease incidence while maintaining ecological integrity [18,19]. Those that adopted a greater number of improved agronomic practices experienced better social and economic outcomes, achieving higher sales prices, increased income, and enhanced overall productivity [20]. Furthermore, avocado production systems with higher levels of innovation adoption showed greater avocado yield and reduced fruit rejection during the selling process, demonstrating the potential for sustainable intensification approaches that enhance both productivity and quality [21].

Despite the emphasis on traditional agronomic factors, there has been limited research exploring the influence of socioeconomic and institutional drivers on avocado production systems [22,23], particularly how these factors interact spatially to create distinct production system typologies within the same geographic region. While spatial approaches have proven successful in analyzing agricultural productivity patterns driven by natural resource endowment, technology adoption, and policy interventions [24], the spatial clustering of avocado production systems based on combined agronomic and socioeconomic characteristics remains unexplored [25,26,27].

Given the complexity of avocado production systems and diverse factors influencing their management, this study typified and characterized avocado production systems in Colombian lowlands of Casanare through spatial autoregressive modeling and clustering techniques to identify distinct typologies presenting different pathways toward sustainable agricultural development. Specifically, this study addresses the following research questions:

(i)

What distinct typologies of Lorena avocado production systems exist in the Colombian lowlands of Casanare?

(ii)

How do agronomic practices and socioeconomic characteristics vary across different production system typologies?

(iii)

What spatial patterns characterize the distribution of avocado production systems in the study region?

The focus on Casanare’s lowland avocado systems is particularly relevant for sustainable agricultural development research as these emerging production areas represent opportunities for implementing sustainable intensification practices from the outset.

2. Materials and Methods

2.1. Study Area and Data Collection

This study was conducted in the municipalities of Tauramena and Monterrey within Casanare department, Colombia (Figure 1). Unlike the traditional growing regions for Lorena avocado in Colombia, such as Antioquia, Tolima, and the Coffee Region [28], avocado production in Casanare represents a recent diversification initiative that is only beginning to be documented in official records [29].

The survey instrument was designed to gather comprehensive data on avocado production systems using a spatial sampling design to achieve geographic coverage [30]. A total of 45 avocado production systems were surveyed between February and December 2023 (Figure 2), selected based on three key inclusion criteria: (i) avocado production systems in the production stage, (ii) farms of varying scales to capture the diversity of production systems, and (iii) farmers with different levels of experience in avocado cultivation. Data collection focused on three main categories of variables: production characteristics (yield, number of trees, plot age, tree yield, and planting distance), agricultural practices (adoption rates of various management practices), and socioeconomic characteristics (farmer age, education level, cooperative membership, market access, income diversification, labor usage, technical assistance, and expansion plans).

2.2. Data Analysis

The collected data were analyzed using descriptive statistics to determine the general characteristics of the surveyed avocado production systems. Following this analysis, the spatial component of the methodology involved three main analytical steps designed to capture both the geographic distribution and the inherent characteristics of avocado production systems: (i) spatial grid creation and data aggregation, (ii) spatial autoregressive modeling to account for geographic dependence in yields, and (iii) geographic clustering analysis combining spatial proximity with production system characteristics.

The spatial grid was designed following established protocols for lattice-based spatial regression modeling, where grid size selection is critical for ensuring spatial connectivity and model validity. The optimal grid size was determined through an iterative process that balanced three key criteria: (i) ensuring spatial continuity by maximizing polygon connectivity across the study region, (ii) guaranteeing that each grid cell contained at least one sampling point to maintain spatial representativeness, and (iii) achieving adequate model performance through validation of spatial regression assumptions including residual normality and independence.

Multiple grid sizes were tested in preliminary analyses, with the 2.2 km resolution selected based on superior model performance indicators and spatial coverage optimization. The grid was generated using QGIS software (version 3.22.4), with the anchor point defined at coordinates −72.85° W, 4.95° N. Figure 2 shows how sampling points are grouped within the 2.2 km grid to determine which points fall within each grid cell, with dark circles representing sampled avocado production systems, light circles showing grid centroids, and rectangles indicating the spatial analysis units.

Once the grid was generated, the sampling values within each polygon were grouped, and the mean and median of avocado tree yields were calculated for each polygon. These average values were assigned to the centroids of each grid cell. Polygons without information were discarded, and in the remaining polygons, centroids were generated for subsequent spatial analysis. This process resulted in 24 spatial units based on the geographical proximity of avocado production systems. The number of polygons was determined by the spatial coverage requirements and the need to maintain a sufficient sample size within each spatial unit while ensuring that the study region was covered with polygons, generating a continuous random field without being separated by a lack of information between adjacent polygons.

For spatial modeling, the taxonomy described by Elhorst [31] was used, in which different models were evaluated using the average yield per polygon unit as the response variable. Multiple spatial regression models were evaluated, including the Spatial Error Model (SEM), Spatial Lag Model (SLM), and Spatial Autoregressive Model (SAR). The SAR model was selected based on three criteria: (i) effective elimination of spatial dependence in residuals, (ii) the statistical significance of estimated parameters, and (iii) a strong relationship between the observed response values and model predictions. An SAR was implemented to account for spatial dependence in avocado yields. This model explicitly considers that observations close in space are related to each other [32], which is particularly relevant for agricultural systems in which neighboring farms often share similar environmental conditions, management practices, and knowledge networks. The final model was a pure autoregressive model with a row-standardized spatial weight matrix using the three nearest neighbors and inverse Euclidean distance. Using the “spautolm” function in the “Spdep” package in R [32], we fitted an autoregressive model in which the yield of each polygon was explained by the yields of neighboring polygons. The model was written as follows:

$$y=\lambda Wy+\epsilon$$

(1)

where y represents the response vector of average yields for each spatial unit, λ represents the autoregressive coefficient measuring the strength of spatial dependence, Wy represents the spatial lag of the response (weighted average of neighboring yields), and ε represents the model residual vector that is expected to be independently and identically distributed under the established conditions of neighbors and standardization of the weight matrix.

The predicted values of the response were extracted from the model, and the assumption of no spatial autocorrelation was evaluated using Moran’s index for the error term. This model predicted the yield of each polygon, incorporating the spatial autocorrelation structure of the data and serving as a key input for subsequent clustering analysis.

The final step of the modeling process involved clustering analysis using both the observed and predicted values from the spatial econometric model. The “ClustGeo” package was employed to group polygons with similar characteristics [33], specifically incorporating spatial constraints in clustering procedures.

Clustering analysis incorporated five key variables: average yield (observed), predicted yield values from the spatial model, longitude coordinates of polygon centroids, latitude coordinates of polygon centroids, and spatial proximity relationships. For cluster calculation, two distance matrices were constructed: a non-spatial variables distance matrix (including average yield and predicted values from the model) and a geographical variables distance matrix (longitude and latitude distances of polygon centroids).

This clustering approach combines both yield-related information and spatial information through an alpha parameter (α) that balances the influence of yield patterns and geographical proximity when forming clusters. The “hclustgeo” function was specifically used to adjust the alpha parameter, optimizing the balance between non-spatial characteristics and spatial constraints. The optimal alpha parameter (α = 0.6) was automatically determined using the ClustGeo algorithm to provide optimal weighting between the spatial component and the feature space related to available dataset information (Figure 3). With this coefficient, the balance between spatial context variables (longitude and latitude distances of polygon centroids) and non-spatial feature variables (average yield and predicted values from the spatial model) is evident in the clustering process.

This analysis resulted in five distinct clusters of avocado production systems, each representing spatial units with similar yield characteristics and geographic proximity. Once the spatial clusters were established, we characterized each group based on three categories of variables: production characteristics, agricultural practices, and socioeconomic characteristics. We calculated central tendency measures (means for continuous variables and percentages for categorical ones) for each cluster to identify distinctive patterns. We evaluated variables such as the average number of trees per farm, yield levels, orchard maturity, technological adoption rates, education levels, institutional connections, and business orientation.

3. Results

3.1. Descriptive Characteristics of Production Systems

The descriptive analysis of the surveyed avocado production systems (n = 45) revealed variation in their production characteristics, agricultural practices, and socioeconomic profiles (

Table 1. Descriptive statistics of production characteristics, agricultural practices, and socioeconomic variables for 45 Lorena avocado production systems in Casanare lowlands, Colombia.

Category and Variables	Description	Value
Production characteristics		Mean ± SD
Yield 1	Annual production (kg)	2908.89 ± 1849.67
Number of trees 1	Trees per farm	321.69 ± 331.00
Tree yield 1	Production per tree (kg tree−1)	17.46 ± 7.22
Planting distance 1	Distance between trees (m)	6.65 ± 1.17
Plot age 2		% Distribution
Initial production (≤5 years)	Farms in early production phase	24.44
Mature production (>5 years)	Farms in production plateau	75.56
Agricultural practices 3		% Adoption
Irrigation	Use of irrigation systems	80.00
Soil analysis	Soil testing practice	73.33
Foliar analysis	Leaf testing practice	17.78
Chemical fertilization	Use of chemical fertilizers	73.33
Organic fertilization	Use of organic fertilizers	93.33
Edaphic fertilization	Soil-applied fertilizers	82.22
Liming	Soil pH management	91.11
Pest control	Pest management practices	97.78
Monitoring	Crop monitoring practices	91.11
Weed control	Weed management practices	97.78
Pruning	Tree pruning practices	91.11
Tool disinfection	Sanitization of tools	84.44
Pruning cut disinfection	Treatment of pruning wounds	84.44
Socioeconomic characteristics
Age 1	Farmer’s age (years)	53.98 ± 11.30
Education level 2		% Distribution
Primary	Up to grade 5	33.33
Secondary	Grades 6–11	26.67
Undergraduate	Technical/university education	35.56
Graduate	Post-graduate education	4.44
Cooperative membership 4	Belonging to cooperatives	31.11
Direct sales 4	Direct market access	80.00
Additional income 4	Other income sources	91.11
Hired labor 4	Use of external labor	95.56
Technical assistance 4	Access to technical support	64.00
Expansion plans 4	Plans to expand production	35.56

¹ Mean ± standard deviation for continuous variables. ² Percentage distribution across categories. ³ Percentage of producers who adopted the practice (coded as 1 = yes, 0 = no). ⁴ Percentage of producers responding “yes” to each characteristic (coded as follows: 1 = yes; 0 = no).

). The average annual yield was 2908.89 ± 1849.67 kg, with farms having 321.69 ± 331.00 trees and a tree yield of 17.46 ± 7.22 kg/tree, on average. The planting distance averaged 6.65 ± 1.17 m, and most farms (76%) were in their mature production phase (>5 years).

Regarding agricultural practices, there was high adoption (>80%) of most management practices, including pest control (97.78%), weed control (97.78%), organic fertilization (93.33%), monitoring (91.11%), pruning (91.11%), and liming (91.11%). However, foliar analysis showed notably low adoption (17.78%), while soil analysis and chemical fertilization were moderately adopted (73.33% each). In terms of socioeconomic characteristics, farmers averaged 53.98 ± 11.30 years of age and had diverse educational backgrounds: 33.33% with primary education, 26.67% with secondary, 35.56% with undergraduate, and 4.44% with graduate education. Although cooperative membership was relatively low (31.11%), most producers had access to direct sales (80.00%), received technical assistance (64.00%), and relied on hired labor (95.56%). A high percentage had additional income sources (91.11%), although only 35.56% planned to expand production.

3.2. Spatial Distribution and Modeling Results

SAR effectively captured the geographic dependence of yield patterns across the region.

Table 2. Parameter estimates and statistical results from the spatial autoregressive model (SAR) applied to average avocado yields across 24 spatial polygons.

Coefficients	Estimate	Std. Error	Statistic(.) 1	Pr(>|.|) 2
Intercept	13.4080	2.6466	z = 5.0663	4.06 × 10−7
Lambda	0.6022	0.1468	LR = 10.3720	0.0013

¹ Statistic column shows z-statistic for intercept (testing if intercept differs from zero) and LR-statistic for lambda (likelihood ratio test for spatial dependence). ² Pr(>|.|) provides p-values for statistical significance testing.

shows the results of the spatial econometric model fit in which a statistically non-zero coefficient for the autoregressive coefficient was evident (λ = 0.6022; p = 0.0013), suggesting the model’s adequacy and how yield is explained by the yield of the established neighborhood structure.

Table 3. Validation tests for spatial autoregressive model assumptions examining residual properties.

Residual Variance	Moran I-Test	Shapiro-Test
26.601	p = 0.2694	p = 0.9451

shows the validation of spatial modeling assumptions. The residual variance was 26.601, and the Moran I-test for spatial autocorrelation in residuals showed no significant spatial dependence (p = 0.2694), confirming that the spatial structure was adequately captured by the model. The Shapiro–Wilk test indicated a normal distribution of residuals (p = 0.9451).

3.3. Geographic Clustering Analysis

Clustering analysis using the ClustGeo package involved the selection of the optimal alpha parameter to balance spatial and non-spatial characteristics. Figure 3 shows the selection process for α = 0.6, which provided the optimal balance between spatial context variables (longitude and latitude distances of polygon centroids) and non-spatial feature variables (average yield and predicted values from the spatial model).

Clustering analysis resulted in five distinct avocado production system clusters, each with specific combinations of production, technological, and socioeconomic characteristics. Figure 4 shows the spatial distribution of these clusters across the study region, revealing clear geographical patterns with certain production system types concentrated in specific areas.

3.4. Production System Typologies

Based on spatial clustering analysis, five distinct types of avocado production systems were identified and characterized (

Table 4. Comparative summary of key characteristics across avocado production system types.

Characteristic	Transitional Traditional (n = 15)	Intensive Technical Management (n = 4)	Experience-Based Traditional (n = 5)	Balanced Management (n = 10)	Comprehensive Technical (n = 11)
Scale and production
Average trees per farm	187.93	162.50	183.40	299.60	644.91
Average yield (kg)	2686.67	3375	3280	2870	2909.09
Mature orchards (>5 y)	86.67%	75%	100%	70%	54.55%
Technical/management aspects
Technological adoption	Moderate	High	Low	Balanced	Very high
Irrigation	73.33%	75%	40%	90%	100%
Soil analysis	66.67%	100%	40%	80%	81.82%
Socioeconomic
Average farmer’s age (years)	52.47	50.00	58.40	55.90	53.73
Predominant education	Mixed higher	Polarized	Primary	Mixed	Higher education
Technical assistance	73.33%	25%	60%	60%	72.73%
Cooperative membership	53.33%	25%	0%	20%	27.27%
Expansion plans	26.67%	50%	60%	10%	54.55%

). Each cluster represents a unique combination of scale, technical adoption, and socioeconomic characteristics.

3.4.1. Transitional Traditional Systems (N = 15)

This cluster represents small-scale production systems with an average of 187.93 trees per farm and mature orchards (86.67% > 5 years) achieving yields of 2686.67 kg. These systems show high adoption of basic practices, including pest control and monitoring (100%) and organic fertilization (93.33%), with moderate adoption of irrigation (73.33%), soil analysis (66.67%), chemical fertilization (66.67%), and edaphic fertilization (66.67%). Foliar analysis adoption was low (26.67%).

This group had a balanced gender distribution (53.33% men), middle-aged farmers (average 52.47 years), and relatively high educational levels (46.67% undergraduate; 40% secondary). It showed the highest cooperative membership rate (53.33%) and significant access to technical assistance (73.33%). Most hired external labor (86.67%) and had low interest in expansion (26.67%).

3.4.2. Intensive Technical Management Systems (N = 4)

This cluster showed the highest performance, with average yields of 3375 kg despite having the smallest orchards (average 162.50 trees). These systems were characterized by 100% adoption of key technical practices, including soil analysis, chemical fertilization, organic fertilization, and liming, with predominantly mature trees (75% > 5 years).

This group consisted of the youngest farmers (average 50 years) with a polarized educational profile (50% undergraduate; 50% primary), with exclusively male farmers and relatively low access to technical assistance (25%). They demonstrated strong commercial orientation with high direct sales (75%) and universal labor hiring (100%). Half of these production systems planned to expand (50%).

3.4.3. Experience-Based Traditional Systems (N = 5)

This cluster consisted of production systems with all orchards in mature production phase (100% > 5 years) achieving high yields (3280 kg). These systems predominantly featured farmers with primary education (80%), were mainly operated by male producers (60%), and had the highest average farmer’s age (58.40 years).

These systems had no cooperative membership (0%) but maintained moderate access to technical assistance (60%) and high direct sales (80%). All systems had additional income sources and hired external labor (100%), with strong expansion plans (60%). Technology adoption was low across multiple areas: irrigation (40%), soil analysis (40%), foliar analysis (40%), and chemical fertilization (40%).

3.4.4. Balanced Management Systems (N = 10)

This cluster represented medium-scale operations averaging 299.60 trees with predominantly mature orchards (70% > 5 years) and moderate yields (2870 kg). These systems featured mixed education levels (40% primary, 40% secondary, and 20% undergraduate) and were predominantly operated by male producers (90%).

The systems showed 100% adoption of organic fertilization, edaphic fertilization, and liming practices, combined with high irrigation adoption (90%). No systems practiced foliar analysis (0%). Cooperative membership was modest (20%), with limited expansion plans (10%).

3.4.5. Comprehensive Technical Systems (N = 11)

This cluster encompassed large-scale operations with an average of 644.91 trees per farm, the most balanced age distribution (54.55% > 5 years), and above-average yields (2909.09 kg). These systems were operated by middle-aged farmers (average 53.73 years) who were the most educated farmers (36.36% undergraduate; 18.18% graduate) and predominantly male producers (72.73%).

They featured high access to technical assistance (72.73%) and the highest direct sales rate (90.91%). Technology adoption was comprehensive, with 100% implementation of irrigation, edaphic fertilization, pest control, and monitoring. All systems hired external labor, and more than half (54.55%) planned to expand.

4. Discussion

The typification and characterization of avocado production systems in this study revealed five distinct groups with varying combinations of technological, productive, and socioeconomic characteristics, demonstrating the heterogeneity of production systems even within the same region. This diversity in production systems suggests the need for targeted development strategies and support mechanisms that recognize multiple pathways to sustainability, as emphasized by Huber et al. [1] and Alvarez et al. [34]

The overall productivity of the analyzed systems (2.9 ton ha⁻¹) fell considerably below both the departmental average of 8 ton ha⁻¹ and the national average of 11.03 ton ha⁻¹ [29]. This substantial gap, combined with the modest average farm size of approximately 322 trees, characterizes the region as dominated by small-scale production systems with generally low productivity. However, the variation in performance across the identified typologies suggests that even within this challenging context, certain combinations of practices and management approaches can lead to improved outcomes. This finding is particularly relevant for sustainable agricultural development initiatives that seek to close yield gaps while minimizing environmental impacts and maintaining social equity.

The observed low productivity likely results from the complex interaction of multiple limiting factors that affect avocado cultivation. Improving avocado yield requires addressing several interconnected aspects, including selecting suitable planting areas that meet specific climatic, topographic, and soil requirements [35]; choosing improved varieties and rootstocks that ensure high-yielding strains with superior phytosanitary quality [36]; implementing adequate water management strategies to optimize crop growth and fruit development [37]; ensuring balanced nutrient management to maintain soil fertility and plant health [38]; establishing optimal planting density and implementing appropriate pruning practices to maximize light interception and tree productivity [39]; enhancing pollination efficiency to improve fruit set and yield potential [40]; and developing comprehensive disease and pest management programs to minimize yield losses [18].

4.1. Multiple Pathways to Sustainable Productivity

In contrast to conventional assumptions about economies of scale in agriculture [41], the highest yields were not associated with the largest operations, suggesting alternative models for sustainable intensification. Despite having the smallest average size (162.5 trees), intensive technical management systems achieved the highest productivity (3375 kg), followed closely by experience-based traditional systems (3280 kg). This pattern, where smaller intensive operations can outperform larger ones, has been observed in other tree crop systems and aligns with the principles of sustainable intensification that emphasize optimizing inputs rather than simply scaling up operations [25]. It also reinforces the findings by Ramírez et al. [18], who found that technological level, rather than size, was a key determinant of avocado farm performance. Similarly, Cáceres-Zambrano et al. [42] demonstrated that technological validation and appropriate technology adoption were more critical than farm size in determining avocado production outcomes.

This productivity paradox demonstrates that technology adoption effectiveness depends on implementation context and farmer capacity rather than simply technology availability, implying multiple viable pathways for agricultural success. The coexistence of different technological approaches, from highly technical to more traditional systems, all achieving relatively similar yields, shows that farmer knowledge and experience can be as valuable as formal technological adoption in achieving sustainable outcomes. Experience-based traditional systems demonstrated that experience-based management can be effective for sustainable production as they achieved the second highest yields despite lower technology adoption. This aligns with the results of Biazin et al. [22], who found that farmer experience could sometimes compensate for lower technological adoption in smallholder systems.

The ability of smaller operations to achieve higher yields may be attributed to their capacity for more precise management and resource optimization, as suggested by Jiménez et al. [9] in their study on data-driven understanding of on-farm yield variation. The intensive adoption of specific practices, such as soil analysis and chemical fertilization, in intensive technical management systems underscores the importance of targeted technological adoption rather than blanket application of all available technologies.

4.2. Comparative Advantages of Production System Typologies

Each identified typology presents distinct advantages and development pathways that reflect different approaches to balancing productivity, sustainability, and economic viability. Intensive technical management systems demonstrate the potential for high productivity through targeted technology adoption, making them suitable models for areas with high market access and technical support availability [12,42]. However, their reliance on external inputs and technical expertise may limit scalability in resource-constrained environments.

Experience-based traditional systems offer valuable insights into low-input sustainable production, achieving high yields through accumulated knowledge and local adaptation strategies. These systems provide important lessons for sustainable intensification approaches that build on traditional ecological knowledge while selectively incorporating modern technologies, demonstrating pathways that emphasize farmer knowledge over external inputs [5,22]. Transitional traditional systems represent an intermediate pathway, showing how gradual technology adoption can be combined with traditional practices, making them particularly relevant for extension programs targeting incremental improvements [1].

Balanced management systems demonstrate the viability of diversified approaches that integrate multiple strategies without specializing in any single direction. This flexibility may provide greater resilience to market and environmental uncertainties, aligning with principles of sustainable agriculture that emphasize adaptability and risk management [6,7]. Comprehensive technical systems, while requiring higher resource investments, showcase the potential for scaling up sustainable production through systematic application of multiple technologies and practices, representing models suitable for areas with strong institutional support and market integration [10,17,20]

4.3. Socioeconomic Characteristics and “Hybrid” Farming Systems

The socioeconomic analysis revealed important patterns that suggest that many avocado farmers in the study area operate as hybrid or part-time farmers, combining agricultural activities with off-farm income sources. This finding is evidenced by the high percentage of producers with additional income sources (91.11%), widespread use of hired labor (95.56%), and relatively high education levels across clusters. These characteristics align with the profile of hybrid farmers who balance agricultural and non-agricultural work to maintain economic stability while pursuing farming activities.

Johansson et al. [43] argued that off-farm income has been recognized as a critical factor enabling continued participation in agriculture as this diversification strategy allows farmers to invest in long-term soil health, adopt environmentally beneficial practices, and maintain farm operations during periods of economic uncertainty, contributing to the overall sustainability of agricultural operations.

The role of cooperative membership and technical assistance showed interesting patterns, with transitional traditional systems having the highest cooperative membership (53.33%), suggesting that these organizations serve as important opportunities for smaller producers to access resources and knowledge. This aligns with the results of Ramirez-Gómez et al. [44], who highlighted the importance of avocado farmer organizations in supporting the adoption of technological and sustainable practices.

4.4. Organic Fertilization and Productivity Trade-Offs

The high adoption of organic fertilization practices (93.33% across all systems) represents a commitment to sustainable soil management but may partially explain the lower productivity observed compared to regional and national averages. While organic fertilizers provide significant environmental benefits, including improved soil structure, enhanced biodiversity, and reduced environmental contamination, research has documented that they can also contribute to yield variability and potentially lower short-term productivity [45,46].

The trade-offs associated with organic fertilization are complex and context-dependent. Studies have shown that organic farming systems often exhibit lower temporal yield stability compared to conventional systems, particularly in fruit crops and during the early years of transition [47]. The challenge for avocado producers in Casanare is optimizing organic fertilization strategies to minimize yield penalties while maximizing sustainability benefits. Research suggests that combining organic and mineral fertilizers in appropriate ratios can help maintain productivity while achieving sustainability goals [48].

4.5. Spatial Patterns and Targeted Interventions

The spatial dimension of our analysis revealed important geographic patterns that have significant implications for sustainable agricultural development strategies. The clustering of similar production systems in specific areas suggests the presence of knowledge diffusion networks and shared environmental conditions that can be leveraged in extension efforts focused on sustainable practices [49,50]. This spatial clustering pattern is consistent with the findings by Anacona et al. [35], who identified specific geographic zones with varying suitability for avocado cultivation in Colombia, emphasizing how environmental and spatial factors influence production system development and sustainability potential.

The geographic concentration of different typologies indicates that local conditions, including biophysical factors and social networks, play important roles in shaping production system characteristics and their sustainability trajectories. This suggests that interventions should be geographically targeted and consider local environmental conditions, social structures, and knowledge networks when promoting sustainable practices.

The identified typologies suggest distinct development trajectories that require differentiated support strategies. Transitional traditional systems represent an important group potentially moving toward more sustainable technical production, while balanced management systems demonstrate a middle path incorporating both traditional ecological knowledge and modern sustainable practices. These different pathways have important implications for extension services and policy support that seek to promote sustainable agricultural intensification [1].

Most systems across all clusters showed high adoption of traditional agricultural practices but varied in their adoption of more specialized practices, such as foliar analysis, suggesting opportunities for targeted technical support and knowledge transfer focused on sustainable intensification approaches. The heterogeneity observed in this typology demonstrates that one-size-fits-all approaches for sustainable agricultural development are ineffective. Instead, support strategies should be tailored to the specific characteristics and needs of each system type [34].

4.6. Implications for Sustainable Development Policy and Future Research

Policy interventions could benefit from this spatial understanding by targeting specific zones for particular types of sustainability-focused support rather than implementing uniform programs across diverse production landscapes. This aligns with the arguments of Notaro et al. [25] and Sekiya et al. [26] that agricultural development strategies should account for spatial variability and diverse management approaches within sustainability frameworks.

The findings suggest that supporting the transition to more sustainable avocado production requires nuanced approaches that recognize the legitimacy of different pathways to sustainability. Transitional traditional systems could benefit from continued cooperative support and a gradual introduction of sustainable technologies, while intensive technical management systems might need support in maintaining their technical efficiency while exploring more environmentally sustainable practices. Experience-based traditional systems could benefit from selective technology adoption that complements their existing successful practices without compromising their sustainability advantages. Comprehensive technical systems could benefit from innovation support and market development that enhances their existing technology sophistication and commercial orientation.

This study acknowledges the limitations that should be considered when interpreting the results. The reliance on spatial clustering within a single region may restrict generalizability to other avocado-producing areas with different environmental and socioeconomic conditions. The relatively small sample size (n = 45) may limit statistical robustness for some analyses, particularly when examining relationships within individual typologies. Additionally, the cross-sectional nature of the data limits causal inferences about long-term sustainability outcomes and technology adoption trajectories. While the identified typologies offer valuable policy insights, practical implementation requires addressing contextual barriers such as uneven technical assistance access and market constraints that may vary significantly across regions and production systems.

Future research should focus on identifying factors that are predictive of productivity through statistical modeling approaches, which can establish causal relationships between specific practices and yield outcomes. Additionally, investigating the long-term sustainability trajectories of hybrid farming systems and their contribution to landscape-level sustainability goals can provide valuable insights for regional development planning, particularly for strengthening Lorena avocado cultivation in Colombia. Research on optimizing organic–mineral fertilizer combinations for tropical fruit systems could help resolve productivity–sustainability trade-offs while maintaining environmental benefits. Longitudinal studies tracking the evolution of these production system typologies over time would provide deeper insights into sustainability transitions and the effectiveness of different intervention strategies.

5. Conclusions

This study addressed the knowledge gap in the spatial characterization of Lorena avocado production systems in Colombian lowlands by identifying five distinct typologies through spatial autoregressive modeling and geographic clustering analysis. The research objective of characterizing production systems to inform sustainable agricultural development strategies was achieved through comprehensive analysis of agronomic practices and socioeconomic characteristics.

The key findings demonstrate that sustainable productivity in smallholder avocado systems follows multiple pathways rather than a single intensification model. Farm size and technology sophistication are less critical than the effective management of practice combinations within each system type, challenging conventional scale-based development approaches. The spatial clustering of similar production systems reveals the importance of geographic context and knowledge networks in shaping agricultural development trajectories.

The identified typologies provide actionable insights for targeted interventions: intensive technical management systems offer models for high-input sustainable production, experience-based traditional systems demonstrate low-input sustainability pathways, while transitional and balanced systems represent intermediate approaches suitable for different resource contexts. The prevalence of hybrid farming systems with off-farm income diversification indicates evolving livelihood strategies that enhance agricultural resilience.

These findings support tailored, spatially targeted interventions that recognize diverse pathways to sustainable intensification rather than uniform development approaches. The methodology provides a replicable framework for characterizing agricultural production systems in other tropical fruit systems and emerging production regions, contributing to evidence-based sustainable agricultural development policy.