Using low-cost NIRS method for helping smallholder to detect nutritional deficiencies and imbalances

Avit Valentin, Flori Albert, Impens Reinout, Aholoukpe Hervé, Vrignon-Brenas Sylvain. 2024. Using low-cost NIRS method for helping smallholder to detect nutritional deficiencies and imbalances. In : Synergies for a resilient future: from knowledge to action. s.l. : s.n., Résumé, 2 p. Congress of the European Society for Agronomy. 18, Rennes, France, 26 Août 2024/30 Août 2024.

Url - autre : https://events.institut-agro.fr/event/1/

Résumé : Lack of control over fertilization is one of the major factors in the yield gap between smallholders and large oil palm plantations (Monzon et al., 2023). The diagnostic tool known as leaf analysis method is used by a large number of plantations to manage their fertilization and relies on annual leaf analysis and specific long-term experiments (Dubos et al., 2022). However, this method is not accessible to smallholders, mainly due to the cost of annual leaf analysis. As a consequence, major imbalances in fertilization are generally observed. The most important nutrients to monitor in this crop are N, P, K and Mg (Woittiez et al., 2017). However, current methods to measure leaf nutrient status are long, costly and require hazardous chemical reagents. Recent developments in near-infrared spectrometry (NIRS) have made it possible to create increasingly low-cost measurement equipment, without the need to transform the leaflets (Prananto et al., 2020). We tested the possibility to use a small portable infrared spectrometer (Nirone S2.2 Evaluation Kit from Spectral Engines) in order to make fertilization management more accessible for smallholders. This spectrometer has a reduced spectral range of 1750-2150nm which has been selected in a prior study testing the correlation between leaflet contents with various spectral ranges. A total of 92 leaflets composite samples were taken from several plots of smallholders and large plantations in West Africa. The plots were located in marginal hydric conditions of Benin and in more favorable conditions of Nigeria, in plantations with different nutritional status, age and plant materials, allowing a large range of variability. Spectral Measurements were taken directly on fresh leaflets on the frond 17, using in classical leaf analysis. Laboratory measurements were got according to the standard methodology of Leaf Analysis. PLS regressions after preprocessing the spectra were used to set up predictive models for the concentration of N, P, K and Mg in the leaves. Goodness of predictions was evaluated by crossvalidation with 10 folds. In our conditions, the measured values (in % of DM) ranged from 1.5 to 3.0 (N), from 0.10 to 0.17 (P), from 0.23 to 1.03 (K) and 0.12 to 0.69 (Mg) which corresponds to leave nutrients content generally observed in oil palms plantations. A satisfying accuracy between measured and predicted nutrients contents was generally observed but depended on the considered nutrient. For instance, N was the better predicted nutrient (RMSECV=0.19). In contrast, Mg was quite poorly predicted (RMSECV = 0.092). In terms of error, the proportion of samples predicted with an error > 20% was lower for N and P (2% and 1%, respectively), compared to K (30%) and Mg (60%). Our results showed that using a portable and cheap infrared spectrometer could be used to easily and quickly predict nutrients contents in a wide-range of nutritional conditions. However, compared to the standard leaf analysis methodology, prediction accuracy is lower especially regarding Mg and K. In a context of advices to smallholders, the trade-off between time, cost and precision is in favor to lower cost even if it means being less precise. This technology appears interesting to (i) detect situations with severe deficiencies and/or imbalances and (ii) educate smallholders to best management practices.

Mots-clés libres : Oil Palm, NIRS, Mineral nutrition

Auteurs et affiliations

• Avit Valentin, CIRAD-PERSYST-UMR ABSys (FRA) ORCID: 0000-0002-1181-6158
• Flori Albert, CIRAD-PERSYST-UMR ABSys (FRA) ORCID: 0000-0002-6962-5792
• Impens Reinout, Presco Plc (NGA)
• Aholoukpe Hervé, INRAB (BEN)
• Vrignon-Brenas Sylvain, CIRAD-PERSYST-UMR ABSys (FRA) ORCID: 0000-0002-5086-3694

Source : Cirad-Agritrop (https://agritrop.cirad.fr/610510/)

Métriques

[ Page générée et mise en cache le 2024-12-16 ]