Fifteen out of 25 experimental trials from the CacaoFIT network are ongoing and continue doing collaborative research with national and international research institutions or universities. The CacaoFIT network has tested a wide range of planting densities and genetic material. Approximately 150 distinct cacao genotypes/clones were assessed in a mixture of 8–10 clones per plot. Across the network, planting density was classified as low (< 625 plants ha⁻¹) in 20% of cases, medium (between 625 and 833 plants ha⁻¹) in another 20%, high (1,111 plants ha⁻¹) in another 45%, and very high (≥1,300 plants ha⁻¹) in the remaining trials (Table 2). Most research trials were managed conventionally (i.e., 55%), defined by the use of synthetic inputs, fertilizers, mechanical weeding, etc. Meanwhile, 20% were managed via organic practices defined by the absence of agrochemicals aside from the application of bioproducts and manual weeding practices. The remaining 25% of trials utilized both conventional and organic management for comparison purposes. Additionally, most research trials (80%) performed soil analysis at the beginning of the experiment and continue to do so every two to three years as a means of monitoring changes in physical and chemical soil properties, as well as the effects on enhancing soil biota.

Most trials (ca. 60%) managed both well-known international clones and national cultivars from selection programs; the remaining 30% grew locally selected elite trees, and only 10% grew seed-based cocoa plantations for comparison purposes (mainly in Bolivia, El Salvador, and Peru). Completed trials run by CATIE in Costa Rica and Panama, as well as managing both seed-based cacao plants from controlled pollination and a set of international clones (Somarriba and Beer, 2011). Interestingly, the experimental trial set up by ECOM and Fundación NicaFrance in Nicaragua is currently testing plants derived from somatic embryogenesis. Most experimental trials (90%) were rain-fed, and only five (the ones located along the dry Pacific coast of Ecuador and El Salvador) were irrigated. Mixed shade canopies were dominant across research trials (70%), while timber-based vs. leguminous shade trees were also tested in 20% of the trials (mainly in Colombia and Central America). Timber-based agroforestry systems were dominant in Honduras, while simple shade canopies or full-sun cacao within the CacaoFIT were less common. The research trials set up by UENF + MARS + CEPLAC tested a wide range of cacao-shade tree combinations, including mixed shade, leguminous trees (Erythrina sp., G. sepium), cacao + coffee (Coffee canephora) + teak (Tectona grandis), cacao + rubber trees (Hevea brasilensis), and other mixtures of cacao with native fruit trees (Anona muricata, Spondias mombin), timber (Schizolobium amazonicum, Tabebuia heptaphylla, C. alliodora, Tabebuia heptaphyllam, Bagassa guianensis, C. guianensis), and palm trees such as Acai (Euterpe oleracea), peach palm (Bactrias gasipaes), coconut (Coconut nuficera), and Brasil nut (Bertholetia excelsa). For a full description of these diverse shade canopies, see Gama-Rodrigues et al. (2021). Surprisingly, only three research trials in the network (FiBL in Bolivia, INIAP in Ecuador, and CRC-Trinidad and Tobago) presented unshaded plots as a control treatment.

The diversity and depth of research domains conducted across the CacaoFIT network varied greatly. Agronomy and agroforestry themes dominated the research agenda of all experimental sites. Regarding agronomy, cacao plant growth and vigor, accumulated yields, and the incidence of pests and diseases were the most common research topics. More complex topics, such as the dynamics of pod production and the effects of pruning (frequency and intensity) on yields, had been recently conducted by a handful of research trials. Regarding agroforestry, the topics documented by nearly 75% of the experimental trials were shade tree growth, generation of goods/products from associated trees (annual crops, timber, fruits, firewood, etc.), and the assessment of canopy cover over time (Figure 1). Shading factors and tree phenology (foliage dynamics) were seldom assessed within the CacaoFIT network (Saj et al., 2013; Magne et al., 2014; Schneider et al., 2017; Armengot et al., 2021; Sauvadet et al., 2021).

Regarding the set of environmental/ecosystem services, most research trials (65%) have documented carbon stocks, sequestration rates, and nutrient cycling, whereas topics such as the abundance/habitat of pollinators, local biodiversity, and soil/micro and macro fauna were assessed to a lesser extent. Other research topics, such as rainfall partitioning and litterfall/decomposition rate, were overlooked across CacaoFIT; so far, the trials led by FiBL in Bolivia, Agrosavia in Colombia, and CEPLAC in Brazil were the only ones that were researched and published on these topics. Climate variables (precipitation, temperature, wind speed, and relative humidity) for most research trials (80%) were gathered by in-site or nearby weather stations, while microclimate variables were rarely measured locally. Again, the research trail from FiBL in Bolivia is leading the way concerning microclimate-shade management-yield relationships.

Concerning the physiological features of cacao plants, namely leaf area index, sap flow, chlorophyll fluorescence, gas exchange (CO₂ assimilation, transpiration, and leaf conductance), and water relationships (leaf water potential and osmotic adjustment), only ~40% of the research trials within CacaoFIT have conducted this set of studies. Physiological measurements were commonly taken by experimental trials in Colombia, Bolivia, and Venezuela, which produced several articles in both English and Spanish. Physiology research topics were almost absent in Mesoamerica and the Caribbean, presumably due to a lack of equipment, instruments, and skilled staff (Ramon E. Jaimez, Universidad Tecnica de Manabi, Ecuador, and personal communication). Finally, regarding the financial performance of cacao farming, annual profitability, and cost/benefit analysis were the most common key financial indicators tested in ~70% of the CacaoFIT network. Except for the study conducted by Ramirez et al. (2001), risk analysis and long-term financial modeling are not fully developed themes. Novel topics such as labor, energy demand, food safety, and lifecycle assessments were assessed only by the trial in Bolivia (Armengot et al., 2016; Pérez-Neira et al., 2020, 2023). Establishment and maintenance costs of both conventional and organic management were researched to a lesser extent. In summary, research gaps across the CacaoFIT network were evident and deserved attention.

In 1986, framed in the cacao and agroforestry research program, FHIA established a network of 36 experimental plots (43 ha in total, ranging in size from 0.5 to 1.5 ha per plot) that combined 12–15 cacao varieties (density=1100 plants ha⁻¹) with 36 timber shade species (29 native species and seven exotic species) aimed at testing the agronomic and agroforestry performance of cacao timber-based agroforestry systems and delivering technical guidelines for cacao farming in humid-lowland Honduras (Figure 2). For over two decades, FHIA and partners have registered monthly data on cocoa yields and by-crops production, costs of agronomic inputs, income from harvested products, and the incidence of pests and diseases. Tree growth parameters (diameter and height) and shade tree phenology features (crown width and shading factors) were recorded annually. Research outcomes generated from this research trial were:

Between 2008 and 2010, FiBL, in partnership with El Ceibo and ECOTOP, set up a network of seven ha of research plots aimed at comparing agroforestry systems and monocultures under both organic and conventional management. A fifth treatment included successional or dynamic agroforestry systems with no external inputs. Gross research plot size was 48 m × 48 m (2,304 m²), while net plots were 24 m × 24 m (576 m²). For all treatments, cacao and plantains were planted at a low density (625 ha⁻¹) (Figure 5). Since 2009, FiBL and partners have regularly registered data on yields of cocoa and by-crops, labor time, costs of agronomic inputs and income from harvested products/goods, tree growth, pests, and diseases, as well as the phenology of cacao trees, soil fertility, and shade canopy management. Research outcomes derived from the SysCom trial were:

In 2008, Agrosavia established a research trial of 1.5 ha in two localities aimed at comparing the performance of international and local cacao clones shaded by native and exotic timber species. The trial was established under a randomized complete block design with nine treatments (nine cocoa genotypes) in a factorial design with three repetitions. Cacao was planted under abarco (Cariniana piryformis) and caucho (Hevea brasiliensis) trees in site #1 and under C piryformis and teca (Tectona grandis) in site #2. The selection of shade tree species responded to local preferences and market potential. Dasometric variables were measured for shade trees, cacao yields, and the incidence of monilia (Moniliophthora roreri) by cacao genotype. Data have been recorded for over 10 years per agroforestry combination, which has yielded several scientific and technical publications. Research outcomes devised from the two medium-term trials were:

In 2014, the University of the Amazonia set up the Macagual Amazon Research Center, Caquetá department, in western Amazonia, comprising 32 ha of cacao-shade tree combinations (AFS) under a randomized complete block design with five replications. In each block, treatments were arranged in strip plots. One strip contained the four AFS, while in the strip perpendicular to the AFS, the clones were randomly planted. The average plot size was 1.5 ha, shaded by several species, including Huito (Genipa americana), Caracoli (Anacardium excelsum), Abarco (Cariniana pyriformis), and Capiron de Vega (Calycophyllum spruceanum). In each block, shade trees were planted at 12 x 12 (70 trees ha⁻¹), and cacao was planted in a north-south direction at 3.5 × 3.5 m (816 plants ha⁻¹). Shade tree species were selected by local farmers based on leaf traits (leaf size, N fixation), canopy traits (crown size and phenology,) and value. The overall goal of the long-term trial is to evaluate the adaptability of both national and international clones shaded by different AFS and under Amazonian conditions. Data collection was carried out from 2018 to 2022, and key research outcomes from this trial were as follows:

Between 2004 and 2010, the CEPLAC Ministry of Agriculture, Livestock, and Food Supply partnered with private actors and organized farmers to implement at least eight different cacao-based agroforestry models across the main production regions in Brazil (MAPA-CEPLAC, 2011) (Figure 13). At the farm level, each cultivation model performed differently in terms of cacao yield, shade canopy products, and, hence, financial revenues to farmers (Table 3). At the landscape level, these cacao cultivation models created an interconnected agroforestry mosaic with natural forests that can be considered climate-smart agriculture, balancing biodiversity protection and commercial production (Schroth et al., 2016a,b). The adoption potential of a given cultivation model is dictated by the productivity and profitability achieved over time. Design features and economic considerations with an emphasis on the yields provided by cacao and the consort of associated shade species are presented elsewhere (Gama-Rodrigues et al., 2021).

The establishment of several cacao-based agroforestry systems by CEPLAC/MAPA considered four key design and management aspects for sustainable agriculture:

CacaoFIT's long-term vision is to generate science-based knowledge and technical guidelines for sustainable cacao cultivation across LAC. However, several research gaps were evident from the CacaoFIT research agenda assessment. Agronomy research questions are being addressed by most research trials (specifically linked to cacao growth and yields and the overall incidence of pest and disease under shaded models), while other key topics are under-examined aspects of cacao cultivation. For instance, the dynamic of pod load vs. cacao plant architecture, the effects of pruning regimes (intensity and frequency) on yields (Orozco-Aguilar et al., 2021; Jaimez et al., 2022; Goudsmit et al., 2023) and the allocation of basal area models (Somarriba and López, 2018a) were found to be seldom researched (Nygren et al., 2013; Heming et al., 2022; Schmidt et al., 2022). Exploring the “crowding effects” of cocoa trees and neighboring trees on the per-plant yield will generate more nuanced advice for best practices in planting density (Wibaux et al., 2017; Cilas and Bastide, 2020; Saj et al., 2023). Below-ground interactions such as fine root dynamics, root volume/biomass and exploration profiles were other topics under-researched within the CacaoFIT network.

Agroforestry-related topics such as the effects of shading factors and tree functional traits (Gagliardi et al., 2020, 2021, 2022, 2023; Isaac et al., 2024) on pathogen dynamics (Leandro-Muñoz et al., 2017; Avelino et al., 2020), rainfall partitioning, and microclimate modificationwere still under-researched topics, especially in comparison to coffee agroforestry trials (Padovan et al., 2015; Abdulai et al., 2020). The SysCom trial in Alto Beni, Bolivia, and the CIMAZ research center in Ecuadorian Amazonia were the only research teams exploring such cacao-shade canopy interactions (Niether et al., 2019, 2020; Armengot et al., 2020, 2023; Hernández-Nuñez et al., 2024). The influence of historical weather and microclimate conditions on yields and the dynamics of pests and diseases is an unexplored yet highly pertinent research issue within the CacaoFIT research agenda. Key environmental services at both farm and landscape levels have been studied by several members of the CacaoFIT network, mostly focused on carbon stock and sequestration potential, litter decomposition, and nutrient cycling. However, soil macrofauna, soil moisture/infiltration, pollinator abundance and diversity, and local/migratory birds were studied to a lesser extent (Toledo-Hernández et al., 2020; Ocampo-Ariza et al., 2024). The restoration potential of shaded cacao plots was not a top-ranked topic in the CacaoFIT research agenda (Schroth et al., 2017; Harvey et al., 2021; Fremout et al., 2022; Bennet et al., 2023).

The study of cacao plant physiology and its interactions with associated trees were minimally explored within the CacaoFIT network. Notable research on this topic has been conducted at the experimental site in Merida, Venezuela, and led by the University de Los Andes (Araque et al., 2012; Ávila-Lovera et al., 2016). Nowadays, the experimental trials located at CIMAZ and Agrosavia, both from Colombia, are levering the topic with experimental and modeling work (Suárez Salazar et al., 2018a; Jaimes-Suárez et al., 2022; Carvalho et al., 2023). The remaining CacaoFIT research trials fall short in this regard, presumably due to the lack of instruments, software, and skilled staff. Topics such as rehabilitation or renovation costs and technical guidelines to do so, although needed in the region (Dalberg, 2015; Somarriba and López, 2018b; Riedel et al., 2019), were seldom evaluated. Although key for decision-making and accessing credits, the financial performance of cacao cultivation models was the least researched or published topic within CacaoFIT. This might be a warning call for all CacaoFIT members to agree on a set of key performance indicators to better communicate results to value chain actors. Finally, farmer outreach was strong and dynamic among a few CacaoFIT members, where several actors were trained, technical publications were delivered, and capacity-building spaces were offered. Large-scale dissemination of research findings from the CacaoFIT trial into farmers' hands and university curricula is a much-needed task of this consortium.

This dynamic context of cacao cultivation in LAC poses social, economic, and environmental challenges to those in charge of knowledge generation. The delivery of cost-effective technical guidelines for thousands of cocoa farmers is essential. In this study, we documented the novel knowledge generated and published by CacaoFIT members, yet we understand that to properly address the industry challenges, only a coordinated effort by all stakeholders can ensure cocoa profitability and sustainability (Shapiro and Rosenquist, 2004). Here, we identified five key actions to strengthen the research agenda, foster collaboration among the CacaoFIT network, seek alliances between CacaoFIT and third parties, better connect with peers in the global south, and deliver mainstream communication and outreach.