| Sumario: | This report synthesizes baseline evidence from 400 surveyed farmers to support the design of a segmented, optimal fertilizer advisory. The analysis combines (i) descriptive profiling of farm size and current nutrient management practices, (ii) adoption modelling for inorganic and organic fertilizer use using logistic regression, (iii) a probability model (random forest) to triangulate the strongest predictors of adoption, (iv) farmer segmentation (k=3 clusters) to translate heterogeneity into actionable advisory groups, and (v) simple scenario simulations to quantify how changes in access (CIS and irrigation) might shift inorganic fertilizer adoption probability. Across the sample, the farming system demonstrates substantial enabling capacity for intensification: irrigation access is high (~86%), livestock ownership is high (~80%), and climate/advisory information channels (CIS) reach roughly 78% of households. Fertility management is already mixed, with similar adoption levels for organic fertilizer use (~61%) and inorganic fertilizer use (~59%). However, mineral fertilizer choices are highly concentrated: DAP (176 records) and CAN (160 records) dominate, while balanced NPK blends and other products appear rarely. This concentration indicates a strong nitrogen–phosphorus focus and a likely gap in potassium and micronutrient coverage, which can limit yield response and nutrient use efficiency (NUE) over time. The adoption models point to distinguish drivers by input type. For inorganic fertilizer use, cultivated land size (land_used) is the most statistically supported predictor (odds ratio 1.15, p=0.0378), indicating that the odds of using mineral fertilizer increase with farm scale. Irrigation shows a positive association (odds ratio 1.66, p=0.0795), suggesting that water security reduces risk and encourages investment in mineral inputs, even though the evidence in this sample is marginal at the 5% level. For organic fertilizer use, livestock ownership is the dominant driver (odds ratio 3.08, p≈2.1e-05), consistent with manure availability as a binding constraint for organic use. Farmer segmentation reveals three advisory-relevant groups: (A) irrigated ISFM smallholders (n=215) who combine organic use with irrigation and strong livestock ownership; (B) irrigated mineral-only users (n=127) who use inorganic inputs but no organics; and (C) rainfed mixed-input farmers (n=58) who lack irrigation and show moderate adoption of both organic and inorganic inputs. These segments imply different risk profiles, different feasible fertilizer packages, and different communication needs. A one-size-fits-all recommendation will underperform relative to a segmented approach. Scenario simulations show modest shifts in predicted inorganic adoption under universal access changes: baseline probability is 0.592; setting CIS access to 100% increases it to 0.602; setting irrigation to 100% increases it to 0.610. The modest magnitude is expected because access is already high for both CIS and irrigation in the sample. The practical implication is that the biggest gains are likely to come from improving recommendation quality (rate, timing, nutrient balance, and ISFM integration), rather than from access expansion alone.
|
|---|