Monitoring vanilla pod quality from Réunion island using near infrared spectroscopy: Assessing moisture and texture during drying

The drying of vanilla pods is a crucial step in the Bourbon processing method. The evolution of dry matter content (DM) and pod firmness (F25 corresponding to the force in Newtons required to cause a 25% strain of the pod’s thickness) were characterized at different stages of the drying process with 150 pods. The increase in dry matter content was not significantly different between the pods at the sun stage (2-4 days of sun drying) and those at the t1 stage (1 to 3 weeks after sweating); DM = 18.4% ± 2.12 vs. DM = 23.9% ± 3.21, respectively. However, between these two stages, the firmness of the pods changed significantly; F25 = 10.6 N ± 3.9 versus F25 = 3.56 N ± 1.82. The increase in DM remained constant thereafter, while the firmness of the pods no longer evolved. These results suggest that the texture modification of the pods is a necessary step for optimal and regular moisture reduction during drying. The study also showed that the visual and manual assessment by processors was generally accurate but could be improved. Predictive models were developed based on the spectral measurements of the pods between 908 and 1676 nm using a portable NIR spectrometer. Partial least squares (PLS) regression with cross-validation showed a high predictive potential for DM (R 2 = 0.97, RMSECV = 2.25%, and RPD = 6.18) and more modest statistics for F25 (R 2 = 0.84, RMSECV = 1.57 N, and RPD = 2.50). Predictions on 50 new pods at different stages demonstrated the robustness of the model for DM content and a lower predictive ability for firmness. A PLS discriminant dnalysis on two classes of pods, “hard” (F25 > 5 N) and “soft” (F25 < 5 N), showed a good discrimination capacity with an accuracy of 0.97 using leave-one-out cross-validation.

Authors

Institutions

Publication Details

Journal
Journal of Near Infrared Spectroscopy
Published
2025-10-21
DOI
https://doi.org/10.1177/09670335251386081
Primary Topic
Biochemical and biochemical processes
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Monitoring vanilla pod quality from Réunion island using near infrared spectroscopy: Assessing moisture and texture during drying

Keshika Mahadeo, Julien Ricci, Hippolyte Kodja, Bertrand Côme et al.
Journal of Near Infrared Spectroscopy
Biochemical and biochemical processes
article

Monitoring vanilla pod quality from Réunion island using near infrared spectroscopy: Assessing moisture and texture during drying

Keshika Mahadeo, Julien Ricci, Hippolyte Kodja, Bertrand Côme, Anne Gauvin‐Bialecki, Fabrice Davrieux, Stéphane Carlier, Guillaume Lalanne‐Tisné
article en

Abstract

The drying of vanilla pods is a crucial step in the Bourbon processing method. The evolution of dry matter content (DM) and pod firmness (F25 corresponding to the force in Newtons required to cause a 25% strain of the pod’s thickness) were characterized at different stages of the drying process with 150 pods. The increase in dry matter content was not significantly different between the pods at the sun stage (2-4 days of sun drying) and those at the t1 stage (1 to 3 weeks after sweating); DM = 18.4% ± 2.12 vs. DM = 23.9% ± 3.21, respectively. However, between these two stages, the firmness of the pods changed significantly; F25 = 10.6 N ± 3.9 versus F25 = 3.56 N ± 1.82. The increase in DM remained constant thereafter, while the firmness of the pods no longer evolved. These results suggest that the texture modification of the pods is a necessary step for optimal and regular moisture reduction during drying. The study also showed that the visual and manual assessment by processors was generally accurate but could be improved. Predictive models were developed based on the spectral measurements of the pods between 908 and 1676 nm using a portable NIR spectrometer. Partial least squares (PLS) regression with cross-validation showed a high predictive potential for DM (R 2 = 0.97, RMSECV = 2.25%, and RPD = 6.18) and more modest statistics for F25 (R 2 = 0.84, RMSECV = 1.57 N, and RPD = 2.50). Predictions on 50 new pods at different stages demonstrated the robustness of the model for DM content and a lower predictive ability for firmness. A PLS discriminant dnalysis on two classes of pods, “hard” (F25 > 5 N) and “soft” (F25 < 5 N), showed a good discrimination capacity with an accuracy of 0.97 using leave-one-out cross-validation.

Journal of Near Infrared SpectroscopyVol. 33(5-6)
Centre de Coopération Internationale en Recherche Agronomique pour le Développement (FR), Université d'Avignon et des Pays de Vaucluse (FR), Université de Montpellier (FR), University of Reunion Island (RE), Écologie Marine Tropicale des Océans Pacifique et Indien (RE)
Openalex Percentile: Top 68%
Biochemical and biochemical processes
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.