Low-cost near infrared spectrometer for discrimination of Malagasy precious woods and their potential substitutes

Abstract Identifying logs without leaves or fruit remains difficult, complicating the control of trade in precious woods, which are often confused with other species. The development of quick and simple tools is therefore essential for field controls, and near-infrared spectroscopy (NIRS) appears to be well suited to this purpose. In recent years, affordable spectrometers have been developed. This paper presents a proof-of-concept study assessing the feasibility and limitations of using low-cost near-infrared spectroscopy for wood species classification, particularly in developing countries where limited resources make affordable solutions highly relevant. Specifically, it evaluates the performance of a low-cost spectrometer (DLP NirScan, 900–1,700 nm, approximately 1,500 USD) in discriminating precious woods from Madagascar and their substitute species across two climatic zones. Twenty-two species were sampled (seven Dalbergia spp, nine Diospyros spp, six substitutes species) with six unaveraged spectra per microcore. Discrimination models were developed using partial least squares-discriminant analysis (PLS-DA). The global model achieved 66 % accuracy, while the local models achieved 70 % and 61 % for the two zones, respectively. Although the accuracy remains limited, their affordability and ease of deployment make them valuable for preliminary screening and decision-making in resource-limited contexts. However, results require confirmation through advanced laboratory analysis.