Mind the leaf anatomy while taking ground truth with portable chlorophyll meters.

Abstract

A wide range of portable chlorophyll meters are increasingly being used to measure leaf chlorophyll content as an indicator of plant performance, providing reference data for remote sensing studies. We tested the effect of leaf anatomy on the relationship between optical assessments of chlorophyll (Chl) against biochemically determined Chl content as a reference. Optical Chl assessments included measurements taken by four chlorophyll meters: three transmittance-based (SPAD-502, Dualex-4 Scientific, and MultispeQ 2.0), one fluorescence-based (CCM-300), and vegetation indices calculated from the 400-2500 nm leaf reflectance acquired using an ASD FieldSpec and a contact plant probe. Three leaf types with different anatomy were included: dorsiventral laminar leaves, grass leaves, and needles. On laminar leaves, all instruments performed well for chlorophyll content estimation (R 2 > 0.80, nRMSE 2 > 0.90, nRMSE 2 = 0.45, nRMSE = 11%) and failed for SPAD. For Norway spruce needles, the relation of CCM-300 values to chlorophyll content was also weak (R 2 = 0.45, nRMSE = 11%). To improve the accuracy of data used for remote sensing algorithm development, we recommend calibration of chlorophyll meter measurements with biochemical assessments, especially for species with anatomy other than laminar dicot leaves. The take-home message is that portable chlorophyll meters perform well for laminar leaves and grasses with wider leaves, however, their accuracy is limited for conifer needles and narrow grass leaves. Species-specific calibrations are necessary to account for anatomical variations, and adjustments in sampling protocols may be required to improve measurement reliability.

Plant phenotyping relevance

携帯型クロロフィルメーターによる葉クロロフィル量推定を、葉の解剖学的差異と生化学測定を基準に比較・検証し、校正とサンプリング改善を提案しているため、植物表現型取得法が中心です。

Code and data availability

The paper's Data availability statement explicitly deposits the study's chlorophyll measurement and trait data in a public Zenodo repository, which is an allowed URL. No separate author analysis code URL is given (analyses were in Matlab/R), so the qualifying asset is the deposited dataset.