Using hyperspectral plant traits linked to photosynthetic efficiency to assess N and P partition

https://doi.org/10.1016/j.isprsjprs.2020.09.006Get rights and content

Abstract

Spatial prediction of photosynthesis requires an understanding of how foliage nitrogen (N) and phosphorus (P) regulate this process and the relationship between these elements and scalable spectral proxies. Hyperspectral imagery has been used to predict important photosynthetic variables such as the maximum rate of carboxylation (Vcmax) and electron transport (Jmax). However, our understanding of how generally applicable these relationships are for plants that are limited by N and P, characterised by respective mass based ratios of N/P ≤ 10 and N/P > 10, is still incomplete as most studies assume N and P co-limit photosynthesis.

Hyperspectral imagery and measurements of photosynthesis were obtained from one-year old Pinus radiata D. Don, grown under a factorial combination of N and P treatments. Using these data, the objectives of this study were to (i) identify whether trees were co-limited or independently limited by N and P, and then use hyperspectral imagery to (ii) partition N and P limited trees, (iii) build models of N and P from a range of hyperspectral indices and (iv) explore links between key plant traits and both Vcmax and Jmax.

Compared to the use of all data, which assumes co-limitation, markedly stronger relationships between N and P and photosynthetic capacity were obtained through splitting data at N/P = 10 (independent limitation) for both Vcmax (R2 = 0.40 vs. 0.59) and Jmax (R2 = 0.38 vs. 0.64). A random forest model was used to accurately partition N from P limited trees and the two main variables used within this model were Photochemical Reflectance Index (PRI) and Solar-Induced Chlorophyll Fluorescence (SIF). Using data from the P limiting phase, the most precise models of P were created using PRI (R2 = 0.75) and SIF (R2 = 0.52). Indices that were proxies for chlorophyll were the most precise predictors of N within the N limiting phase but strong positive relationships were also evident between N and both PRI (R2 = 0.83) and SIF (R2 = 0.57). Through their correlations with N and P, there were strong positive relationships between both SIF, PRI and Vcmax (R2 = 0.78 and 0.83, respectively) and Jmax (R2 = 0.80 and 0.83, respectively) that were generalisable across both N and P limiting ranges. These results suggest that quantified SIF and PRI from hyperspectral images may have greater precision and generality for predicting both foliage nutrition and biochemical limitations to photosynthesis than other widely used hyperspectral indices.

Introduction

The prediction of foliage nutrient concentration across broad spatial scale has considerable utility for assessing health and productivity of ecosystems (McNeil et al., 2007a, McNeil et al., 2007b, Townsend et al., 2003). Significant growth responses within forest stands have been noted through the application of fertiliser, mainly in the form of nitrogen (N) and phosphorus (P) (Albaugh et al., 2003, Allen et al., 2005, Liechty and Fristoe, 2013). However, determining the optimal application of fertiliser is often very difficult as nutritional deficiencies vary widely across the landscape (Campion, 2008, Fox et al., 2007).

Foliage nutrition is also a key determinant of rates of carbon assimilation. Many studies have investigated the use of N and P to account for variation in photosynthetic capacity (Walker et al., 2014) and the importance of these elements on the photosynthesis process is well documented. Nitrogen is a major component of Rubisco (Niinemets and Tenhunen, 1997) and P has an impact on many important aspects of photosynthesis including membrane solubility, ATP, and NADPH production (Marschner, 1995, Taiz et al., 2015). Given these strong links it is important to understand how N and P regulate photosynthesis when developing models of these elements that will be subsequently used to predict photosynthesis.

Hyperspectral imagery has been successfully used to predict foliar concentrations of N for a wide range of broadleaf and coniferous species at both the leaf level and canopy level (for an overview see Hill et al., 2019, Watt et al., 2019). Nitrogen shows marked absorption features in both the visible near infrared (VNIR) and short-wave infrared (SWIR) ranges (Curran, 1989, Kokaly, 2001, Kokaly et al., 2009) and models are able to capitalise on these features to predict N (Watt et al., 2019). At the leaf level the precision of models of N has been found to range widely (R2 range of 0.37–0.99) but generally N has been predicted with moderate to high levels of precision, with mean R2 of 0.77, recorded across a wide range of species (Asner and Martin, 2008, Asner et al., 2011, Curran et al., 2001, Dechant et al., 2017, Gillon et al., 1999, Luther and Carroll, 1999, Masaitis et al., 2014, Petisco et al., 2005, Schlerf et al., 2010, Serbin et al., 2014, Stein et al., 2014, Tsay et al., 1982, Wang et al., 2018, Wang et al., 2015, Yoder and Pettigrew-Crosby, 1995). Despite the difficulties in scaling to the canopy, moderate to strong predictions of N have also been made at this level with values of R2 ranging from 0.48 to 0.98 (Coops et al., 2003, Huang et al., 2004, Knyazikhin et al., 2013, Martin et al., 2008, Martin et al., 2018a, Martin et al., 2018b, Ollinger et al., 2008, Singh et al., 2015, Smith et al., 2003).

Although many studies have developed models of P, the lack of spectral absorption features associated with P is likely to limit their generality as these relationships are indirect. Leaf level hyperspectral data have generally been found to predict P empirically with moderate to high precision (mean R2 of 0.74) using proxies correlated with P, with R2 varying from 0.32 to 0.95 (Asner and Martin, 2008, Asner et al., 2011, Curran et al., 2001, Gillon et al., 1999, Masaitis et al., 2014, Petisco et al., 2005, Stein et al., 2014). However, as P does not directly absorb energy in the shortwave spectrum, these predictions of P are actually due to the correlations with N found under most conditions (Asner and Martin, 2008, Gillon et al., 1999, Porder et al., 2005). As a consequence, these models may not be as robust when applied to conditions where ratios of N/P deviate from typical ranges. Development of models of P using datasets with little correlation between N and P are likely to reveal the true precision between hyperspectral-based traits and P, likely resulting in predictions that have greater generality.

Although little research considers whether N and P co-limit or independently limit photosynthesis, this is a key assumption that is likely to affect model precision, generality and applicability. Datasets used generally assume that photosynthesis is co-limited by N and P as both elements are predicted across their complete range. Although it has not been widely investigated, a number of studies suggest that N and P independently limit growth and photosynthesis in many species (Bown et al., 2007, Domingues et al., 2010, Ingestad, 1971, Ingestad, 1979, Ingestad and Lund, 1986). The premise underlying this research is that a mass based N/P ratio of 10 is optimal with values of N/P ≤ 10 leading to N limitations and N/P > 10 resulting in P limitations (Aerts and Chapin, 2000, Marschner, 1995, Reich and Schoettle, 1988). Previous research supports this suggestion through showing that N and P independently influence the key biochemical limitations to photosynthesis that include the maximum rate of carboxylation (Vcmax) and electron transport (Jmax) (Domingues et al. 2010) and that a stoichiometric ratio of 10 can be used to partition N from P limitations for these variables (Bown et al. 2007). If photosynthesis is independently regulated by N and P it follows that Vcmax and Jmax will exhibit positive relationships with both elements within their respective limiting ranges and these relationships will be weaker when they are constructed using the entire dataset under the assumption of co-limitation.

The assumption of independent limitations has important implications for the development of foliage nutrition models. In order to identify the type of required nutrient additions it would be useful to be able to partition N from P limitations at scales ranging from the tree to the forest level. Following this partitioning, the use of N and P models developed using data from their respective limiting ranges could be applied to estimate the severity of any deficiency and impact on photosynthesis within these two ranges. This approach may provide a means of improving spatial accuracy when characterising the type and extent of nutrient deficiencies. Understanding the nature of relationships, within each limiting range, is also likely to provide considerable insight into the mechanistic link between plant functional traits quantified from hyperspectral imagery and foliage nutrition and, as a consequence, growth and photosynthesis. Despite this, we are unaware of any research that has used this approach for developing models of N and P for tree species.

Traditional methods used to track changes in plant nutrition from remote sensing have most often targeted chlorophyll content (Ca+b) as chlorophyll and other pigments such as carotenoids, xanthophyll and anthocyanins are important indicators of plant photosynthetic status (Baret et al., 2007, Evans, 1989, Yoder and Pettigrew-Crosby, 1995). Nitrogen is a major component of chlorophyll, and nitrogen and chlorophyll deficiencies are directly related to reductions in photosynthetic rates (Evans, 1989). It therefore follows that Ca+b has been the focus of remote sensing research as a proxy for nitrogen within agriculture and forestry (Baret et al., 2007, Yoder and Pettigrew-Crosby, 1995).

Remote sensing research carried out in the 1980s identified the red-edge and green spectral regions as potential targets for estimating Ca+b that were linked to nitrogen content (Carter, 1994, Gitelson and Merzlyak, 1996, Rock et al., 1988). Further research developed specific narrow-band hyperspectral indices (Haboudane et al., 2002) including the red-edge Chlorophyll Index (Zarco-Tejada et al., 2001). Combined indices were also developed such as the Transformed Chlorophyll Absorption in Reflectance Index, TCARI (Haboudane et al., 2002) normalized by the Optimized Soil-Adjusted Vegetation Index, OSAVI (Rondeaux et al., 1996) to form the TCARI/OSAVI proxy for chlorophyll and nitrogen.

In addition to Ca+b, recent research has identified other pigments, such as xanthophylls, that are more dynamically related to rapid changes in photosynthesis and are potentially more useful for tracking nutritional impacts on photosynthesis. The changes observed in the green spectral region through the Photochemical Reflectance Index (PRI) (Gamon et al., 1992) have been demonstrated to be linked to the xanthophyll cycle, and this index has been successfully used to predict photosynthetic rate (Drolet et al., 2008, Fuentes et al., 2006, Gamon et al., 1997, Guo and Trotter, 2004, Hilker et al., 2008, Middleton et al., 2009, Nichol et al., 2000, Penuelas et al., 1995b, Stylinski et al., 2000) and the photosynthetic response of plants to a range of stresses (Buddenbaum et al., 2015, Dobrowski et al., 2005, Hernández-Clemente et al., 2011, Scholten et al., 2019, Suarez et al., 2008).

During the last 50 years (see review by Mohammed et al., 2019) considerable research has demonstrated the utility and feasibility of Solar-Induced Chlorophyll Fluorescence (SIF) in predicting photosynthetic activity at both the leaf and the canopy scales from a range of remote sensing platforms (Cendrero-Mateo et al., 2015, Zarco-Tejada et al., 2013, Zarco-Tejada et al., 2016). Given the strong relationship between chlorophyll pigments, the xanthophyll dynamics (PRI) and photosynthesis (through SIF) it also follows that these indicators might be significantly related to N and P as these two elements are key determinants of photosynthesis rate.

Pinus radiata D. Don (radiata pine) is the most widely planted plantation species within the southern hemisphere (Lewis and Ferguson, 1993). Over 4.1 M ha of this species has been established in New Zealand, Chile and Australia, where the species comprises, respectively, 90% (NZFOA, 2018), 62% (Salas et al., 2016) and 39% (Downham and Gavran, 2019) of the total plantation area. Pinus radiata frequently suffers from nutrient limitations, particularly during mid-rotation when nutrient demand often exceeds supply. The key elements that limit productivity of P. radiata are N and P (Watt et al., 2005). A shortage of these elements can result in significant reductions in growth (Raison and Myers, 1992, Sheriff et al., 1986) and also limits the key processes that control the rate of photosynthesis (Bown et al., 2009).

In this study, measurements of hyperspectral imagery, photosynthesis and foliage nutrition were taken from an experiment that included a factorial combination of N and P treatments applied to P. radiata. Using these data, the initial objective of this research was to (i) identify whether trees were co-limited or independently limited by N and P, and then use hyperspectral imagery to (ii) partition N and P limited trees, (iii) build models of N and P from a range of hyperspectral indices and (iv) explore links between key plant traits and both Vcmax and Jmax.

Section snippets

Experimental set up

The experiment was undertaken within the Scion nursery, located in Rotorua, New Zealand. A total of 120 P. radiata seedlings were transplanted into 15 L pots during October 2018. The medium into which the plants were transplanted consisted of a mixture of perlite and vermiculite which are silica-based products without any nutritional content. Plants were grown in a thermostatically controlled greenhouse where temperature in spring fluctuated between 10 and 24 °C during the day and between 10

Foliar nutrition

The applied treatments resulted in a wide range in N and P (Fig. 1). Values of N ranged from 0.41 to 2.12% when expressed on a mass basis and 11.1–49.0 µg cm−2 on an area basis while P varied from respectively 0.052–0.330% and 1.33–7.66 µg cm−2. The relationship between N and P was insignificant when data was expressed on a mass (R2 = 0.0073; P = 0.52) or area basis (R2 = 0.0006; P = 0.85). Of the 60 plants, 43 were categorised as N limited while 17 were P limited and the separation of these

Discussion

This study advances our understanding of the functional relationships between SIF, PRI and both foliage nutrition and photosynthesis. The strongest relationships between N and P and both Vcmax and Jmax were developed through splitting the data at N/P = 10 suggesting that photosynthesis is independently limited by these elements. The model that was used to accurately partition N from P limited trees demonstrated the importance of PRI and SIF in separating these two groups. Within each of these

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

We are grateful to Kendra Newick who assisted with the preparation of the nutrient solutions. The project was partly funded through the Resilient Forests programme, which is funded through Scion SSIF as well as the Forest Grower’s Levy Trust. Funding was also received from the National Institute for Forest Products Innovation (Project Number NIF073-1819), which comprised contributions from the Australian Government, Australasian Forestry Companies and South Australian and Tasmanian State

References (159)

  • A. Damm et al.

    Far-red sun-induced chlorophyll fluorescence shows ecosystem-specific relationships to gross primary production: An assessment based on observational and modeling approaches

    Remote Sens. Environ.

    (2015)
  • C. Daughtry et al.

    Estimating corn leaf chlorophyll concentration from leaf and canopy reflectance

    Remote Sens. Environ.

    (2000)
  • B. Dechant et al.

    Estimation of photosynthesis traits from leaf reflectance spectra: correlation to nitrogen content as the dominant mechanism

    Remote Sens. Environ.

    (2017)
  • S.Z. Dobrowski et al.

    Simple reflectance indices track heat and water stress-induced changes in steady-state chlorophyll fluorescence at the canopy scale

    Remote Sens. Environ.

    (2005)
  • G.G. Drolet et al.

    Regional mapping of gross light-use efficiency using MODIS spectral indices

    Remote Sens. Environ.

    (2008)
  • D.A. Fuentes et al.

    Mapping carbon and water vapor fluxes in a chaparral ecosystem using vegetation indices derived from AVIRIS

    Remote Sens. Environ.

    (2006)
  • J. Gamon et al.

    A narrow-waveband spectral index that tracks diurnal changes in photosynthetic efficiency

    Remote Sens. Environ.

    (1992)
  • M.F. Garbulsky et al.

    The photochemical reflectance index (PRI) and the remote sensing of leaf, canopy and ecosystem radiation use efficiencies: A review and meta-analysis

    Remote Sens. Environ.

    (2011)
  • A. Gitelson et al.

    Spectral reflectance changes associated with autumn senescence of Aesculus hippocastanum L. and Acer platanoides L. leaves. Spectral features and relation to chlorophyll estimation

    J. Plant Physiol.

    (1994)
  • A.A. Gitelson et al.

    The chlorophyll fluorescence ratio F 735/F 700 as an accurate measure of the chlorophyll content in plants

    Remote Sens. Environ.

    (1999)
  • A.A. Gitelson et al.

    Relationships between leaf chlorophyll content and spectral reflectance and algorithms for non-destructive chlorophyll assessment in higher plant leaves

    J. Plant Physiol.

    (2003)
  • A.A. Gitelson et al.

    Use of a green channel in remote sensing of global vegetation from EOS-MODIS

    Remote Sens. Environ.

    (1996)
  • A.A. Gitelson et al.

    Novel algorithms for remote estimation of vegetation fraction

    Remote Sens. Environ.

    (2002)
  • A.A. Gitelson et al.

    Signature analysis of leaf reflectance spectra: algorithm development for remote sensing of chlorophyll

    J. Plant Physiol.

    (1996)
  • D. Haboudane et al.

    Hyperspectral vegetation indices and novel algorithms for predicting green LAI of crop canopies: Modeling and validation in the context of precision agriculture

    Remote Sens. Environ.

    (2004)
  • D. Haboudane et al.

    Integrated narrow-band vegetation indices for prediction of crop chlorophyll content for application to precision agriculture

    Remote Sens. Environ.

    (2002)
  • R. Hernández-Clemente et al.

    Assessing structural effects on PRI for stress detection in conifer forests

    Remote Sens. Environ.

    (2011)
  • Z. Huang et al.

    Estimating foliage nitrogen concentration from HYMAP data using continuum removal analysis

    Remote Sens. Environ.

    (2004)
  • A. Huete et al.

    A comparison of vegetation indices over a global set of TM images for EOS-MODIS

    Remote Sens. Environ.

    (1997)
  • T. Kattenborn et al.

    Advantages of retrieving pigment content [μg/cm2] versus concentration [%] from canopy reflectance

    Remote Sens. Environ.

    (2019)
  • R.F. Kokaly

    Investigating a physical basis for spectroscopic estimates of leaf nitrogen concentration

    Remote Sens. Environ.

    (2001)
  • R.F. Kokaly et al.

    Characterizing canopy biochemistry from imaging spectroscopy and its application to ecosystem studies

    Remote Sens. Environ.

    (2009)
  • H. Lichtenthaler et al.

    Detection of vegetation stress via a new high resolution fluorescence imaging system

    J. Plant Physiol.

    (1996)
  • J.E. Luther et al.

    Development of an index of balsam fir vigor by foliar spectral reflectance

    Remote Sens. Environ.

    (1999)
  • A. Maccioni et al.

    New vegetation indices for remote measurement of chlorophylls based on leaf directional reflectance spectra

    J. Photochem. Photobiol., B

    (2001)
  • M.E. Martin et al.

    A generalizable method for remote sensing of canopy nitrogen across a wide range of forest ecosystems

    Remote Sens. Environ.

    (2008)
  • R.E. Martin et al.

    Remote measurement of canopy water content in giant sequoias (Sequoiadendron giganteum) during drought

    For. Ecol. Manage.

    (2018)
  • M. Meroni et al.

    Remote sensing of solar-induced chlorophyll fluorescence: Review of methods and applications

    Remote Sens. Environ.

    (2009)
  • G.H. Mohammed et al.

    Remote sensing of solar-induced chlorophyll fluorescence (SIF) in vegetation: 50 years of progress

    Remote Sens. Environ.

    (2019)
  • A.M. Mouazen et al.

    Comparison among principal component, partial least squares and back propagation neural network analyses for accuracy of measurement of selected soil properties with visible and near infrared spectroscopy

    Geoderma

    (2010)
  • C.J. Nichol et al.

    Remote sensing of photosynthetic-light-use efficiency of boreal forest

    Agric. For. Meteorol.

    (2000)
  • H. Aasen et al.

    Quantitative remote sensing at ultra-high resolution with UAV spectroscopy: a review of sensor technology, measurement procedures, and data correction workflows

    Remote Sens.

    (2018)
  • R. Aerts et al.

    The mineral nutrition of wild plants revisited: A re-evaluation of processes and patterns

    Adv. Ecol. Res.

    (2000)
  • M.O. Akinwande et al.

    Variance inflation factor: as a condition for the inclusion of suppressor variable (s) in regression analysis

    Open J. Stat.

    (2015)
  • T.J. Albaugh et al.

    Vegetation control and fertilization in midrotation Pinus taeda stands in the southeastern United States

    Ann. Forest Sci.

    (2003)
  • H.L. Allen et al.

    What is ahead for intensive pine plantation silviculture in the south?

    South. J. Appl. For.

    (2005)
  • F. Baret et al.

    Quantification of plant stress using remote sensing observations and crop models: the case of nitrogen management

    J. Exp. Bot.

    (2007)
  • G.A. Blackburn

    Spectral indices for estimating photosynthetic pigment concentrations: a test using senescent tree leaves

    Int. J. Remote Sens.

    (1998)
  • L.C. Blakemore et al.

    Methods for chemical analysis of soils

    NZ Soil Bureau Scientific Report

    (1987)
  • H.E. Bown et al.

    Partititioning concurrent influences of nitrogen and phosphorus supply on photosynthetic model parameters of Pinus radiata

    Tree Physiol.

    (2007)
  • Cited by (0)

    View full text