A methodological framework for evaluating cacao agroforestry systems at the landscape scale through the Main Agroecological Structure (MAS)

Main Purpose of the Research The primary purpose of this research is to develop specific methodological tools and indicators for analyzing the Main Agroecological Structure (MAS) in Cocoa Agroforestry Systems (AFS). This methodological framework was designed for use by researchers, professionals, or technicians engaged in the structural analysis of production systems within a landscape context. The scope and nature of data collection are empirical and methodological, carried out in the southwestern region of Colombia, specifically in the municipalities of Mocoa, Orito, Puerto Guzmán, and Villagarzón, in the department of Putumayo. * Data Collection Period: Data were collected between 2023 and 2024. * Study Sample: Data collection was conducted on 44 randomly selected farms. * Units of Analysis: The farms were selected from a larger group previously categorized into five typologies of cocoa producers. Main Thematic Areas and Special Contents The study focuses on the structural analysis of agroecosystems, addressing the complexity of cocoa production and the need for conservation strategies in tropical landscapes (land-blending strategy). The main thematic areas are covered through ten sub-indicators that compose the Main Agroecological Structure (MAS): 1. Landscape Structure and Connectivity (Geospatial and Biophysical Themes): Includes Farm–Landscape Ecological Correspondence (FLEC), which assesses whether the farm replicates or alters the surrounding forest cover pattern; and Connection with the Main Ecological Structure of the Landscape (CMELS), which measures the proximity between the farm centroid and the nearest forest edge. 2. Biodiversity Connectors (Biophysical Themes): Includes the Extension of External Connectors (EEC) (live fences along farm boundaries) and the Extension of Internal Connectors (EIC) (rows of vegetation within the agroecosystem). 3. Diversity and Land Use (Biophysical Themes): Includes the Forest Tree Species Richness (FTSR) in cocoa AFS; the Richness of Cocoa Genetic Materials (RCGM), assessing the number of cocoa clones; and Soil Use and Conservation (USC), which measures the proportional distribution of landscape units with permanent cover (including forest, fallow, and cocoa AFS). 4. Agronomic Management (Cultural Practices): Evaluated through a survey, including Agronomic Practices (AP) (fertilization, pest, and weed management); Perception, Awareness, and Knowledge (PAK) regarding the use and management of agrobiodiversity; and the Level of Action Capacity (CA) to establish or improve the MAS. Methodology:1. Study Area and Sample Primary data collection was conducted between 2023 and 2024 in the southwestern region of Colombia, encompassing the municipalities of Mocoa, Orito, Puerto Guzmán, and Villagarzón, located in the department of Putumayo. Main Sample: Data were collected from 44 randomly selected farms. Reinforced Geospatial Sample: Data from the polygons of 44 farms were incorporated to strengthen the development of spatial sub-indicators: Farm–Landscape Ecological Correspondence (FLEC) and Connection with the Main Ecological Structure of the Landscape (CMELS). 2.Collection of Geospatial and Biophysical Data Geographical and biophysical information from the farms was obtained through previous research and advanced image processing techniques: Field Georeferencing: The perimeters of farms and the polygons of crops, pastures, fallows, and conservation areas were obtained by walking along the boundaries together with the farmers, using a Garmin GPSMAP® 30S. Data for the landscape units used in the Soil Use and Conservation (USC) indicator, as well as for the External (EEC) and Internal (EIC) connectors. Topological Analysis: A topological correction process was applied to avoid overlaps between polygons, correcting minor intersections and removing nested polygons within larger ones. Forest Cover Estimation: The Brandt et al. (2023) product was used, derived from multi-temporal convolutional neural network models applied to Sentinel imagery (10 m) and radar data. Data gaps were filled using the Hansen et al. (2013) forest cover product (30 m). Enhanced Area of Influence (AOI) Method: To calculate FLEC, the elliptical method was proposed and selected for defining the farm’s Area of Influence (AOI). This approach proved to be the most reliable, showing a standard deviation of only 1.08 in the AOI-to-farm area ratio, compared to 28.5 for the circular method. This result allowed the establishment of a constant (k) of 38.3 for AOI definition. Calculation of CMELS (Connection with the Ecological Structure): The procedure was implemented in Python 3.11.8. The minimum Euclidean distance was calculated from each farm’s centroid to the nearest forest edge, including directional distances toward the north, south, east, and west. The calculation was refined iteratively with 0.1-meter steps to improve accuracy. 3. Collection of Biophysical Diversity Data Species richness was...