When the protocol encounters the limitations of the field: Capitalizing on experiences from the MIP (2024-2025)

In any research project, drafting a protocol is a fundamental step. Clear objectives, structured methodology, defined indicators, and a precise timeline give the impression of a controlled framework. Yet once the research is deployed in the field, unforeseen parameters can disrupt the initial plan: social constraints (participant availability), equipment failures, logistical difficulties in accessing sites, organizational readjustments, or non-climatic agronomic factors such as pests, differences in farming practices, and soil quality. This experience capitalization document aims precisely to draw lessons from five cases where a gap emerged between the theoretical protocol and field reality during MIP experiments conducted in 2024–2025 across three regions of Madagascar.

Case 1 — Pak Choi at Ambohidray (Alaotra Mangoro Region, 2024): The study focused on minimal irrigation using soil moisture probes. The protocol deliberately excluded any fertilizer input, based on the hypothesis that optimal water management alone would be sufficient to ensure germination. Result: 0% germination across 108 m². A mandatory restart of the experiment resulted in an estimated loss of 450,000 Ariary (seeds, labor, and transportation costs) and a two-month delay in the academic timeline of the lead student, Mr. Ericson Rakotoarisoa, generating significant stress and additional logistical expenses. Lesson learned: Soil nutrient availability has a major influence on germination and must be factored into the protocol design from the outset.

Case 2 — Green Bean at Talata Volonondry (July 2025): This study tested a solar-powered drip irrigation system. The solar pump failed to adequately fill the storage tank due to insufficient flow rate. Despite an alternative controlled irrigation solution implemented after a few days, the yield obtained was only 324 kg/ha, well below projected performance. Lesson learned: The actual energy output capacity of the system must be thoroughly field-tested under real operating conditions prior to full deployment.

Case 3 — Potato at Bakaro (Analamanga Region, April 2025): The study examined the effect of biological inputs on potato production. The protocol was scrupulously followed in all respects — planting calendar, plant density, irrigation scheduling, and crop monitoring. However, a sudden severe hailstorm destroyed the experimental plot, resulting in an estimated 80% loss of expected yield and rendering the dataset non-representative. Lesson learned: Extreme climatic risk assessment must be integrated as a mandatory component of initial experimental planning.

Case 4 — Carrot at Tsarasaotra (Amoron'i Mania Region, 2024): The study examined the water requirements of carrots using soil moisture probes. Certified seeds purchased from a SOC-approved supplier were used. Yet nearly all of the harvested carrots proved unmarketable and unfit for consumption due to a varietal issue known as "male carrots" (forking or bolting defect). The applicability of the results was severely limited, reducing the agronomic and socioeconomic relevance of the findings for the partner farming communities. Lesson learned: Seed quality must be systematically verified before use, regardless of supplier certification.

Supplementary Case — Watermelon at Ambohidray: Certified seeds were used for watermelon trials. However, the seed lot turned out to be mixed, producing two different varieties on the same plot and making comparative analysis unreliable. Nevertheless, plot T3 (soil moisture controlled at 65% relative humidity via probe) yielded results close to the control plot (daily hand-watering), representing a partial positive outcome for the minimal irrigation study.

Comparative analysis of the five experimental situations reveals three recurring patterns: hypotheses applied directly under real field conditions without a prior pilot phase (4 out of 5 cases); external risks — climatic or technical — underestimated in the initial planning (3 out of 5 cases); and assumed input quality without prior germination testing or varietal homogeneity checks (2 out of 5 cases). These patterns underscore the methodological need to introduce a systematic pilot phase before any full-scale experimentation.

In response to these situations, the team made several structural decisions: adjusting the methodology based on the constraints encountered; revising the project timeline without compromising the scientific objectives; systematically documenting all protocol modifications; integrating a comprehensive agronomic baseline analysis prior to experimentation (soil fertility assessment, minimum nutritional requirements); including a climatic risk analysis section in each protocol; verifying seed batches through prior germination and varietal homogeneity testing; and documenting experimental failures as legitimate scientific results. The goal was not to abandon scientific rigor, but to intelligently adapt the methodological framework to observed reality.

A mandatory checklist must be completed before setting up any protocol under farmer field conditions: soil fertility verification (N-P-K analysis or agronomic soil assessment); equipment testing under real operating conditions (not based solely on manufacturer specifications); climatic risk identification through historical weather data analysis covering at least five years; seed germination testing on a minimum sample of 50 seeds; definition of a contingency plan (Plan B) for each major identified risk; allocation of a failure management budget representing 10 to 20% of the total project budget; and a flexible academic calendar with a built-in safety buffer of one to two additional months.

This evolution translates concretely into a transition from rigid protocols — characterized by an estimated experimental failure rate of approximately 60% and schedule overruns in 40% of trials — to adaptive protocols that systematically incorporate risk management and include financial and scheduling safety margins.

These experiences demonstrate that a rigorous protocol does not protect against biological limitations (soil fertility), technical limitations (actual output capacity of a solar pump), extreme climatic events, or seed traceability issues. As the MIP project coordinator emphasizes: "Research conducted with and for farmers takes place under real, on-farm conditions. The risk level is always significantly higher, and unforeseen events are frequent. Unlike controlled laboratory settings or experimental research stations where nearly all variables are managed, farmer plots expose researchers to a complex and unpredictable reality that must be anticipated."

Scientific quality does not reside in the rigid application of an unchanging plan, but in the capacity to adjust the research approach while maintaining methodological coherence. Experimental failure is not a lack of rigor — it is often the indicator of a variable that was not accounted for in the initial model. These experiences now constitute a valuable knowledge base for improving the methodological quality of future agricultural research.