Identifying Tree Species in the Amazon with Spectroscopy

The Amazon Rainforest is one of the most biodiverse ecosystems in the world. Containing 10% of all species on Earth, it is an ecosystem rich in life and natural beauty (1). Because of its biodiversity, researchers are keen to identify the various tree species in the Amazon.

A recent study published in the journal Global Ecology and Conservation explored this topic. In this study, the research team, comprised of researchers from the National Institute for Amazonian Research (INPA), demonstrated how full-range spectroscopy can accurately identify tree species (2).

Why is species identification in the Amazon difficult?

Species identification in the Amazon is notorious for being difficult because of the region’s extraordinary biodiversity (1). Another challenge is that there are a limited number of trained taxonomists and striking morphological similarities between species, which often leads to identification errors that ripple through ecological studies, conservation planning, and forest management (2).

What did this study explore?

The research team evaluated full-range spectroscopy across 26 abundant tree species found in three major Amazonian ecosystems, which included upland forest, white-sand ecosystems, and floodplain forest. By collecting spectral data from three types of tissues (outer bark, inner bark, and fresh leaves), the team assessed how well each tissue contributed to accurate species identification (2). They tested their spectral models using linear discriminant analysis (LDA) with two cross-validation methods, as well as external validation using samples collected more than 300 kilometers from the original study sites (2).

Out of the three tissue types, fresh leaves produced the highest discrimination accuracy at 98%, followed closely by inner bark at 97% and outer bark at 86% (2). Importantly, these high accuracy rates held across all ecosystems, demonstrating that the method is resilient to variations in habitat, soil composition, and forest structure.

The study also pinpointed which regions of the light spectrum are most informative for distinguishing species. For outer bark, wavelengths in the SWIR I region (1300–1900 nm) were most important, while inner bark and fresh leaves relied on a combination of visible wavelengths (400–700 nm) and SWIR I (2). These findings provide a foundation for optimizing future devices and analytical methods for rapid field identification.

An important outcome from this study was that a model was trained using spectral samples from all three ecosystems. This model proved that species identification does not require separate models for each forest type, which is an essential finding for scaling up spectroscopy across the Amazon (2).

External validation further underscored the technique’s potential. Even when applied to species from distant sites, fresh leaves and inner bark were classified with high accuracy, demonstrating that spectroscopy can perform reliably beyond local conditions (2). However, tests involving distant populations must account for unresolved taxonomic issues (2).

What are the key takeaways from this research?

The main takeaway from this study is that the ASD FieldSpec 4, the instrument used in the study, may have the power to conduct analysis in the Amazon Rainforest, but there are practical issues regarding its adoption. For one, the device is both expensive and cumbersome, requiring a laptop and external battery (1). This setup is not practical for the Amazon Rainforest. To properly analyze and evaluate tree crowns and stems in this region, smaller spectrometers are going to be more ideal.

Therefore, the next steps are clear. The development of portable spectrometers, a trend that is ongoing in this space, needs to continue so that they can be built to maintain the high spectral resolution and range necessary to achieve high prediction accuracy (1). A move toward portable, affordable spectrometers combined with robust spectral models could revolutionize biodiversity inventories, enable more reliable forest monitoring, improve species-level quality control, and support conservation initiatives across the Amazon (1).

References

1. World Wildlife Fund, Top facts about the Amazon. WWF.org. Available at: https://www.wwf.org.uk/learn/fascinating-facts/amazon (accessed 2025-12-09).
2. Hadlich, H. L.; Schongart, J.; Wittmann, F.; et al. Exploring the potential of field spectroscopy for tree species identification in different Amazonian forest ecosystems. Glob. Ecol. Conserv. 2025, 64, e03970. DOI: 10.1016/j.gecco.2025.e03970