Mapping phosphorus concentration in Mediterranean forests using different remote-sensing methods

Abstract

Mineral nutrition is essential for optimal plant growth. Phosphorus (P) is a relatively small component of leaf dry weight, with a concentration in plant foliage of less than 1%. Despite its low concentration, P is an essential element in plants, mainly used for energy transfer. Mapping P concentration using traditional methods is expensive and usually limited to a small area; it is timeconsuming and covers only a few plant individuals or species. In this study, we demonstrate the use of remote-sensing (RS) data acquired from the feld and airborne hyperspectral sensors to predict and map the P concentration in leaves of different woody Mediterranean plant species. Comprehensive field work included leaf sampling, laboratory analyses, and spectral measurements using a visible, near-infrared and shortwave-infrared (VIS-NIR-SWIR) field spectrometer. Using different spectral configurations, we built accurate models to predict P concentration in leaf samples. The models were built using a NIR data analysis technique with the data mining software PARACUDA II. This software allowed us to identify the correlative wavelengths for P-bearing molecules in selected woody Mediterranean plant species. The hyperspectralbased model for leaf P concentration was extracted from the reflectance data acquired using a manned aircraft carrying a hyperspectral sensor (Specim AisaFenix 1 K). The model gave a reliable correlation between points extracted from the hyperspectral image and samples measured in the field. We believe that the methodology used in this study will help forest ecologists better understand the concentration of P in the foliage of woody Mediterranean plant species.

Disentangling the effects of phosphorus, nitrogen and species identity on the vegetation reflectance spectrum

“…Differences in vegetation architecture may have contributed to the relatively strong species and community identity effects on the reflectance spectrum, but may also have overruled the effects of biochemistry on the vegetation reflectance spectrum. Most studies trying to predict vegetation nutrient content using vegetation spectra did not consider community composition and hence do not discriminate between species (but see Hellmann et al, 2015; Girard et al, 2020; Mandelmilch et al, 2021). Given that nutrient‐rich soils harbour different species communities compared with nutrient‐poor soils (DeCock et al, 2022) and our finding that species or community identity is a strong predictor for reflectance, it is clear that empirical studies in semi‐natural communities have difficulties measuring vegetation nutrient content directly, but rather make an indirect estimation of vegetation nutrient content via species composition.…”

“…potassium, calcium, magnesium) are potentially located in the short-wave infrared region (SWIR; Lim et al, 2017 and references therein). However, these indicative wavelengths for phosphorus are suggested to be indirect indicators because they have been identified previously as direct absorption features of nitrogen-containing molecules (Porder et al, 2005;Homolová et al, 2013;Mandelmilch et al, 2021). Experiments that are designed to disentangle the effects of phosphorus versus nitrogen may thus provide an important way forward to explore possible indicative wavelengths for phosphorus.…”