Edaphic homologous zones and digital tools as a basis for sustainable soil management in potato growing areas in Colombia – Nature

Report on Edaphic Zoning for Sustainable Potato Production in Colombia

Executive Summary

This report details a study on the spatial heterogeneity of soil properties across Colombia’s primary potato-growing regions. The objective was to delineate homogeneous edaphic zones to facilitate site-specific management, thereby advancing key Sustainable Development Goals (SDGs), including SDG 2 (Zero Hunger), SDG 12 (Responsible Consumption and Production), and SDG 15 (Life on Land). Through the analysis of 3,137 georeferenced soil samples, three distinct edaphic clusters were identified using an unsupervised K-means clustering algorithm. The findings indicate that 59% of the mapped area is highly suitable for potato cultivation. A digital decision-support platform was developed, integrating an evidence-based fertilization recommendation system and a suitability model. This framework promotes efficient input use, reduces environmental impact, and enhances the resilience of potato production systems, offering a scalable model for sustainable agriculture in the Andean region.

1.0 Introduction

1.1 Background: Potato Cultivation and Sustainable Development

Potato (Solanum tuberosum L.) is the fourth most important agricultural commodity globally and a staple food crucial for achieving SDG 2 (Zero Hunger). In Colombia, potato cultivation is concentrated in high-altitude Andean regions characterized by highly variable soils derived from volcanic ash. Intensive agricultural practices in these areas pose significant risks of soil degradation, undermining long-term productivity and threatening terrestrial ecosystems, a core concern of SDG 15 (Life on Land). A comprehensive understanding of soil spatial variability is therefore essential for developing sustainable management practices that align with SDG 12 (Responsible Consumption and Production).

1.2 Study Objectives

The primary objective of this study was to quantify and map the spatial variability of key soil physicochemical properties in Colombia’s potato-producing landscapes. The specific aims were to:

1. Delineate homogeneous edaphic management zones (MZs) using multivariate and machine learning techniques.
2. Characterize the identified zones based on soil fertility indicators, particularly pH, organic matter, phosphorus, and iron.
3. Develop a framework for site-specific fertilization recommendations to optimize nutrient use and minimize environmental impact.
4. Integrate these findings into a digital decision-support tool to promote sustainable agricultural practices among stakeholders.

2.0 Methodology

2.1 Data Compilation and Physicochemical Analysis

The study utilized a database of 3,137 georeferenced soil samples collected from potato production plots across seven departments in Colombia. Samples were collected at a depth of 0-30 cm prior to tillage and fertilization. A comprehensive analysis was performed to determine key soil attributes:

• Physical Properties: Textural class (sand, silt, clay percentages) determined by the Bouyoucos method.
• Chemical Properties: pH, organic matter (OM), cation exchange capacity (CEC), total nitrogen (N), available phosphorus (P), and various macro- and micronutrients.

Data quality was ensured through a rigorous cleaning process involving the removal of outliers and the application of an autoencoder model for anomaly detection, establishing a final dataset of 2,867 records.

2.2 Delineation of Homologous Edaphic Zones

An unsupervised K-means clustering algorithm was applied to the multivariate soil data to delineate homogeneous edaphic zones. The optimal number of clusters was determined to be three, based on internal validation metrics including the average silhouette coefficient and the Calinski-Harabasz index. This data-driven zoning provides a scientific basis for precision agriculture, a key strategy for advancing SDG 12 by tailoring inputs to specific soil conditions, thereby reducing waste and pollution.

2.3 Productivity Modeling and Sustainable Management Tools

A multi-focus modeling approach was implemented to support sustainable management decisions:

1. Productivity Analysis: Potato yield data was analyzed in relation to the identified soil and climatic clusters to understand genotype-by-environment interactions, contributing to the productivity and food security targets of SDG 2.
2. Genetic Algorithm for Nutrient Optimization: A genetic algorithm was adapted to optimize fertilizer recommendations for N, P, K, Ca, and Mg. This tool aims to maximize yield while minimizing nutrient inputs, directly supporting the efficient resource use mandated by SDG 12.
3. Digital Decision-Support Platform: The models and recommendation systems were integrated into ‘SOLANA’, an open-access web platform designed to provide actionable guidance to producers and extension agents.

3.0 Results and Analysis

3.1 Characterization of Edaphic Clusters

The K-means analysis successfully identified three distinct edaphic clusters with clear contrasts in soil properties, providing a foundation for targeted interventions that protect soil health as per SDG 15.

• Cluster 1: Predominantly found in Boyacá, Cauca, and Santander. Characterized by silty loam textures, strongly acidic pH (average 5.1), and higher dispersion in organic matter and nitrogen content. This cluster showed low suitability for CEC, Ca, Mg, and Cu.
• Cluster 2: Most frequent in Cundinamarca and Norte de Santander. Exhibited a different textural pattern, with a higher frequency of sandy loam soils. This cluster showed low suitability for sulfur, boron, and copper.
• Cluster 3: Dominant in Antioquia and Nariño. Characterized by silty loam textures and significantly higher average phosphorus levels (five times higher than Cluster 2). This cluster demonstrated the highest average potato yield (33.81 t ha⁻¹).

3.2 Soil Suitability and Productivity

Based on crop nutritional requirements, 59% of the total mapped area was classified as having high suitability for potato cultivation, with the remaining 41% classified as medium suitability. Yield analysis revealed that the combination of Soil Cluster 2 and Climatic Cluster 4 produced the highest average yield (35.27 t ha⁻¹), underscoring the importance of integrated soil and climate management for achieving the food production goals of SDG 2.

3.3 Spatial Modeling and Digital Tool Implementation

A Gradient Boosting Machine (GBM) model was successfully trained to spatialize the edaphic clusters, achieving an overall accuracy of 83.8%. The most influential variables for spatial prediction were soil organic carbon, silt content, sand content, and pH. The outputs, including the nutrient recommendation system, were integrated into the ‘SOLANA’ digital platform. This tool operationalizes the study’s findings, empowering farmers with data-driven insights to adopt more sustainable production patterns (SDG 12).

4.0 Conclusion and Recommendations for Sustainable Development

This study provides a robust, data-driven framework for sustainable soil management in Colombia’s potato sector, with direct applications for achieving multiple Sustainable Development Goals. The delineation of three distinct edaphic zones enables a transition from uniform management to precision agriculture, which is fundamental for enhancing resource efficiency (SDG 12), boosting productivity (SDG 2), and preventing land degradation (SDG 15).

Recommendations:

1. Promote Adoption of the ‘SOLANA’ Platform: Widespread use of the digital decision-support tool should be encouraged to facilitate the implementation of site-specific management practices.
2. Implement Cluster-Specific Nutrient Management: Fertilization strategies should be tailored to the specific chemical properties of each edaphic cluster to optimize nutrient uptake and reduce environmental runoff.
3. Invest in Continuous Data Collection: Ongoing monitoring of soil health and productivity is necessary to refine the models and update recommendations, ensuring the long-term sustainability of the potato production system.
4. Scale the Framework: The reproducible methodology presented in this report should be scaled to other Andean tuber systems and regions to advance sustainable agriculture on a broader scale.

Analysis of Sustainable Development Goals (SDGs) in the Article

1. Which SDGs are addressed or connected to the issues highlighted in the article?

1. SDG 2: Zero Hunger

   • The article focuses on improving the productivity and sustainability of potato cultivation, which is described as a “staple food in many countries” and the “fourth most important agricultural product.” The entire study is aimed at enhancing food production by providing tools for “optimal yield” and “sustainable crop production,” directly contributing to food security.

2. SDG 12: Responsible Consumption and Production

   • The research promotes sustainable production patterns by enabling “more efficient input use” and reducing “environmental burdens.” The development of an “evidence-based fertilization recommendation system” is designed to prevent the overuse of chemical inputs, aligning with the goal of managing natural resources efficiently.

3. SDG 15: Life on Land

   • The central theme is “sustainable land management” and mitigating “soil degradation” caused by intensive agricultural practices. By mapping soil properties and delineating management zones, the study provides a framework to “contribute to mitigating soil degradation” and improve soil health, which is crucial for protecting terrestrial ecosystems.

4. SDG 9: Industry, Innovation, and Infrastructure

   • The study heavily relies on innovation and technology. It employs “data science techniques,” “machine learning (ML),” and a “genetic algorithm” to create a “digital decision-support platform” (‘SOLANA’). This work enhances scientific research and upgrades the technological capabilities of Colombia’s agricultural sector.

2. What specific targets under those SDGs can be identified based on the article’s content?

1. Target 2.3: Double the agricultural productivity and incomes of small-scale food producers.

   • The study aims to improve potato yield through site-specific management. The development of a system that provides “precise fertilizer dose recommendations for optimal yield” and predicts “potential yield (t ha⁻¹)” directly supports the goal of increasing agricultural productivity for potato farmers in Colombia.

2. Target 2.4: Ensure sustainable food production systems and implement resilient agricultural practices.

   • The article’s framework is designed to create “improved resilience of potato production systems” by aligning agronomic decisions with soil variability. It promotes practices that “help ensure sustainable crop production and environmental sustainability” and “progressively improve land and soil quality.”

3. Target 12.2: Achieve the sustainable management and efficient use of natural resources.

   • The nutrient recommendation system is a key output that enables “more efficient input use.” By optimizing fertilizer application based on soil data, the approach reduces waste and minimizes the environmental impact of agriculture, directly contributing to the efficient use of natural resources like soil and nutrients.

4. Target 15.3: Combat desertification, restore degraded land and soil, and achieve a land degradation-neutral world.

   • The article identifies that intensive potato farming can lead to “soil degradation, with losses in resilience, productivity, sustainability.” The proposed zoning and management guidance provide a direct strategy to combat this by promoting practices that mitigate degradation and support “sustainable soil management.”

5. Target 9.5: Enhance scientific research and upgrade technological capabilities.

   • The entire methodological approach, which integrates “geostatistics, geographic information systems (GIS) and machine learning (ML),” and culminates in an “open-access platform” for farmers, is a clear example of enhancing scientific research and applying advanced technology to an industrial sector (agriculture) to foster innovation.

3. Are there any indicators mentioned or implied in the article that can be used to measure progress towards the identified targets?

1. Potato Yield (t ha⁻¹)

   • This is a direct indicator for Target 2.3. The article explicitly analyzes yield data, stating, “Soil Cluster 3 presents the highest average yield of 33.81 t ha⁻¹.” Tracking this metric over time would measure progress in agricultural productivity.

2. Soil Physicochemical Properties

   • These serve as indicators for Target 15.3. The study measures numerous properties like “pH, organic matter (OM), available phosphorus (P), and exchangeable aluminum (Al).” Monitoring these variables over time can assess whether land management practices are improving soil quality or mitigating degradation. The classification of soil “aptitude” (high, medium, low) also acts as a composite indicator of soil health.

3. Fertilizer Application Doses (kg ha⁻¹)

   • This is an implied indicator for Target 12.2. The nutrient recommendation system generates “calculation of the doses (kg ha⁻¹) for each of the evaluated elements.” Measuring the reduction in fertilizer use while maintaining or increasing yields would demonstrate more efficient resource management.

4. Adoption of the Digital Decision-Support Platform

   • This is an implied indicator for Target 9.5. The article mentions the platform ‘SOLANA’ is for the “use of producers and extensionists.” The number of users, the geographic area covered by its recommendations, and its integration into farming practices would measure the uptake of this agricultural innovation.

4. Summary Table of SDGs, Targets, and Indicators

Source: nature.com