Plant Physiology Preview. Published on November 15, 2016, as DOI:10.1104/pp.16.01447
1
Short Title: High-throughput phenotyping of maize leaf traits
2
Corresponding author: Elizabeth Ainsworth, USDA ARS Global Change and Photosynthesis
3
Research Unit, 1201 W. Gregory Dr., Urbana, IL 61801, USA
4
5
Title: High-throughput phenotyping of maize leaf physiological and biochemical traits
6
using hyperspectral reflectance
7
8
Craig R. Yendrek1, Tiago Tomaz1, Christopher M. Montes1,2, Youyuan Cao1, 3, Alison M.
9
Morse4,5, Patrick J. Brown1,6, Lauren M. McIntyre4,5, Andrew D.B. Leakey1,2, Elizabeth A.
10
Ainsworth1,2,7
11
12
1
13
Urbana, IL, 61801
14
2
Department of Plant Biology, University of Illinois at Urbana Champaign, Urbana, IL, 61801
15
3
Plant Protection Department, Fujian Agriculture and Forestry University, Fuzhou, China,
16
350002
17
4
18
32610
19
5
Genetics Institute, University of Florida, Gainesville, FL, 32610
20
6
Department of Crop Sciences, University of Illinois at Urbana Champaign, Urbana, IL, 61801
21
7
USDA ARS, Global Change and Photosynthesis Research Unit, Urbana, IL
Carl R. Woese Institute for Genomic Biology, University of Illinois at Urbana Champaign,
Department of Molecular Genetics and Microbiology, University of Florida, Gainesville, FL,
22
23
One Sentence Summary. Partial least squares regression modeling of leaf reflectance spectra
24
provides a high-throughput and accurate approach to phenotyping photosynthesis and leaf
25
biochemistry in maize.
26
27
Footnotes. Author contributions: C.R.Y., E.A.A., P.J.B., L.M.M., A.D.B.L. conceived of and
28
designed the original research plans; C.R.Y., T.T., C.M.M. performed most of the experiments;
29
Y.C. provided technical assistance to C.R.Y.; C.R.Y., A.M.M., L.M.M., E.A.A. analyzed the
30
data; C.R.Y., E.A.A. wrote the article with contributions from all of the authors.
1
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
Copyright 2016 by the American Society of Plant Biologists
31
Funding information: This work is supported by the National Science Foundation under Grant
32
No. (PGR-1238030). T.T. was supported by a Fulbright Western Australia Scholarship.
33
34
Corresponding author email: lisa.ainsworth@ars.usda.gov
2
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
35
ABSTRACT
36
High-throughput, non-invasive field phenotyping has revealed genetic variation in crop
37
morphological, developmental and agronomic traits, but rapid measurements of the underlying
38
physiological and biochemical traits are needed to fully understand genetic variation in plant-
39
environment interactions. This study tested the application of leaf hyperspectral reflectance (λ =
40
500 nm – 2400 nm) as a high-throughput phenotyping approach for rapid and accurate
41
assessment of leaf photosynthetic and biochemical traits in maize. Leaf traits were measured
42
with standard wet-lab and gas exchange approaches alongside measurements of leaf reflectance.
43
Partial least squares regression (PLSR) was used to develop a measure of leaf chlorophyll
44
content, N content, sucrose content, specific leaf area (SLA), maximum rate of PEP
45
carboxylation (Vp,max), [CO2]-saturated rate of photosynthesis (Vmax), and leaf oxygen radical
46
absorbance capacity (ORAC) from leaf reflectance spectra. PLSR models accurately predicted
47
five out of seven traits, and were more accurate than previously used simple spectral indices for
48
leaf chlorophyll, N content and SLA. Correlations among leaf traits and statistical inferences
49
about differences among genotypes and treatments were similar for measured and modeled data.
50
The hyperspectral reflectance approach to phenotyping was dramatically faster than traditional
51
measurements, enabling over 1000 rows to be phenotyped during midday hours over just 2 to 4
52
days, and offers a non-destructive method for accurately assessing physiological and biochemical
53
trait responses to environmental stress.
3
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
54
55
INTRODUCTION
Meeting future global food demand is projected to require a doubling of agricultural
56
output by 2050 (Tilman et al., 2011). Climate change will exacerbate this challenge by
57
intensifying the exposure of field crops to abiotic stress conditions, including rising temperature,
58
increased drought, flooding and air pollution (Christensen et al., 2007). In order to more rapidly
59
adapt crops to the future growing environment, diverse germplasm must be meaningfully
60
assessed in field conditions. However, our capacity to phenotype thousands of crop genotypes in
61
a field environment is constrained by available high-throughput phenotyping platforms (Furbank
62
and Tester, 2011; Araus and Cairns, 2014), and our ability to adapt crops to global climate
63
change requires facilities that accommodate hundreds of genotypes in a realistic environment
64
(Ainsworth et al., 2008).
65
This study provides proof of concept for one approach for high-throughput phenotyping
66
of leaf physiological responses to global change using hyperspectral reflectance spectra of
67
diverse maize inbred and hybrid lines grown at ambient and elevated ozone concentrations ([O3])
68
in the field. Hyperspectral sensors capture electromagnetic radiation reflected from vegetation in
69
the visible (400-700 nm), near infrared (700-1300 nm) and short-wavelength infrared (1400-
70
3000 nm) regions, which contain information about leaf physiological status, including pigments,
71
structural constituents of biomass and water content (Curan, 1989; Martin and Aber, 1997;
72
Penuelas and Filella, 1998). Vegetation spectra can be captured in a few seconds and universal
73
indices have been proposed to estimate leaf foliar traits such as chlorophyll content, N content
74
and photosynthetic radiation use efficiency from reflectance data (Gamon et al., 1997; le Maire
75
et al., 2004; Herrmann et al., 2010). Predictive models using partial least squares regression
76
(PLSR) approaches to scale spectral data also have been applied to estimate crop yield (Weber et
77
al., 2012) and to estimate leaf features including isotopic ratios (Richardson and Reeves, 2005),
78
the ratio of leaf mass to leaf area (Asner and Martin, 2008; Asner et al., 2011), carbohydrate
79
content of plant tissues (Dreccer et al., 2014; Asner and Martin 2015) and the maximum
80
photosynthetic capacity of C3 plants (Serbin et al., 2012; Ainsworth et al., 2014). Variation in
81
foliar reflectance at different wavelengths in the spectrum is specific to variation in different,
82
chemical and structural components of leaves (Curran, 1989; Serbin et al., 2014). Therefore,
83
analysis of foliar reflectance spectra has the potential to very rapidly assess multiple
84
physiological and biochemical traits from a single measurement.
4
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
85
The use of hyperspectral reflectance as a high-throughput phenotyping tool in agricultural
86
research is promising (Weber et al., 2012; Araus and Cairns, 2014), but questions remain
87
regarding limitations of the technique. Reflectance spectroscopy has been used to successfully
88
distinguish broad-scale spatial and temporal variation in leaf foliar traits and photosynthetic
89
metabolism across species in diverse agroecosystems (Serbin et al., 2015), tropical ecosystems
90
(Asner and Martin, 2008; Asner et al. 2011), and northern temperate and boreal forests (Serbin et
91
al., 2014). In this study we test the utility of leaf hyperspectral reflectance as a high-throughput
92
method for phenotyping diverse germplasm subject to abiotic stress in maize. Maize is one of the
93
world’s most important crops and also represents a model for species with C4 photosynthesis.
94
The relationship of hyperspectral reflectance with C4 leaf traits has the potential to be distinct
95
from that of C3 species due to the significantly different leaf anatomy, biochemistry and
96
physiology of the two photosynthetic types.
97
We grew diverse maize genotypes in the field under ambient and elevated ozone
98
concentrations ([O3]) using Free Air Concentration Enrichment (FACE). Ozone is an air
99
pollutant that enters leaves through the stomata where it rapidly forms other reactive oxygen
100
species that cause damage to biochemical constituents and physiological processes, ultimately
101
leading to reductions in leaf longevity and yield (Fiscus et al., 2005; Ainsworth et al., 2012). Past
102
efforts to detect the genetic variation in O3 tolerance using mapping populations have scored
103
visible leaf damage or plant biomass (Kim et al., 2004; Frei et al. 2008; Broché et al., 2010;
104
Street et al., 2011; Ueda et al. 2015), but have not quantified the many other physiological
105
changes that ultimately determine the impact of O3 on crop yield. These include reduced
106
photosynthetic capacity, reduced leaf chlorophyll and nitrogen (N) content, accelerated
107
senescence, increased antioxidant capacity, altered carbohydrate and metabolite content,
108
decreased specific leaf area (SLA) and increased rates of respiration (Fiscus et al., 2005; Leitao
109
et al. 2007; Dizengremel et al. 2008; Betzelberger et al., 2012; Gillepsie et al., 2012). The
110
logistics and cost of employing traditional phenotyping methods to quantify these physiological
111
and biochemical markers of stress are prohibitive at the scale needed to screen large populations,
112
and so hyperspectral reflectance offers a promising non-destructive, rapid and high through-put
113
alternative. However, this alternative approach has yet to be validated.
114
115
In order to test the throughput of this approach in the field, we collected reflectance
spectra from diverse inbred and hybrid maize lines grown in ambient and elevated [O3] in
5
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
116
replicated plots across a 16 ha field over three consecutive growing seasons, as well as
117
greenhouse-grown maize (inbred B73) given optimal and limiting N supply. This range of
118
conditions and material was developed to ensure that a wide range of values were measured.
119
These same leaves were sampled using “gold standard” techniques of gas exchange and wet
120
laboratory biochemical approaches to quantify traits relevant to leaf stress physiology, including
121
chlorophyll content, N content, specific leaf area (SLA), maximum rate of PEP carboxylation
122
(Vp,max), [CO2]-saturated rate of photosynthesis (Vmax), leaf oxygen radical absorbance capacity
123
(ORAC) and sucrose content. The two datasets were used successfully to build predictive PLSR
124
models for 5 of 7 traits and to demonstrate that similar conclusions about genetic and treatment
125
variation in leaf traits were drawn from both measured and modeled data.
126
127
RESULTS
128
Predicting Physiological Traits from Reflectance Spectra
129
PLSR models using the leave-one-out cross validation approach predicted leaf
130
biochemical and physiological traits with varying degrees of accuracy. The predictive ability was
131
strongest for chlorophyll (R2=0.81-0.85; Fig. 1a, b) and N content (R2=0.92-0.96; Fig. 1c, d),
132
good for SLA (R2=0.68-0.78; Fig. 1e, f), Vmax (R2=0.56-0.65; Fig. 1g, h) and sucrose content
133
(R2=0.62; Fig. 2a), and non-significant for Vp,max (Fig. 1i, j) and ORAC (Fig. 2b). In general,
134
models created with spectra collected from leaves exposed to multiple growth environments (i.e.,
135
greenhouse and field, low and high N, ambient and elevated [O3]) captured a greater proportion
136
of trait variation than models built from just field-grown plants. For example, a stronger
137
correlation between measured and model-predicted chlorophyll values was observed with the
138
combined field and greenhouse model (R2 = 0.85; Fig. 1b) compared to the field-only model (R2
139
= 0.81; Fig. 1a). Similarly, the correlations for N content and Vmax were stronger using both
140
greenhouse and field data. In contrast, inclusion of data from greenhouse-grown maize plants did
141
not improve the SLA model (Fig. 1f). No greenhouse samples were collected for ORAC and
142
sucrose quantification; therefore, models consisted only of data from the field (Fig. 2). PLSR
143
model coefficients for each trait are provided in Supplemental Table 1.
144
As an additional test of PLSR model stability, 20% of the measured values used for
145
training the original PLSR models were randomly selected and removed (as a holdout dataset). A
146
second set of PLSR models was built using the remaining 80% of the original training data (80%
6
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
147
model). The parameter estimates generated by the 80% model were very similar to the
148
parameters estimates from the original, full model (Supplemental Table 2). In addition, the 80%
149
model accurately estimated the measured values of the holdout dataset that were removed prior
150
to model building (Supplemental Table 2).
151
In order to test the similarity in relationships among traits for measured and PLSR-
152
modeled data, pairwise correlations among traits were examined (Fig. 3). In both measured and
153
PLSR-modeled datasets, the correlations were consistent, demonstrating that the relationships
154
among the underlying measured traits were accurately reflected by the PLSR models. N content
155
and Vmax were positively correlated, and there were also significant correlations between Chl
156
content and N content and Vmax (Fig. 3). N content was negatively correlated to Suc content and
157
Chl content was negatively correlated to SLA (Fig. 3).
158
PLSR models were able to distinguish the same effects as measured traits. Using
159
genotypes B73, Mo17 and the B73 x Mo17 hybrid, analysis of variance (ANOVA) models
160
estimating the effects of genotype, O3 treatment and the interaction had consistent inferences for
161
3 of 4 successful PLSR models (Chl, N, and Vmax; Supplemental Table 3; Supplemental Fig. 1).
162
For SLA the p value for the O3 effect on measured data was 0.012 and the p value for the PLSR-
163
modeled data was 0.095 (Supplemental Table 3).
164
To assess the relative contributions of different wavelengths across the measured spectra
165
as drivers of a given PLSR model, variable importance in projection (VIP) scores for each trait
166
model were compared. The chlorophyll model was very sensitive to variation in reflection of
167
visible wavelengths, with major peaks at 554 nm and 719 nm and minor contributions from the
168
short-wavelength infrared regions (Fig. 4a). The N content model was built with wavelengths
169
between 1500 and 2400 nm based on the dominant absorption features of N (Curran, 1989;
170
Serbin et al., 2014), and was sensitive to variation in reflectance throughout that region of the
171
spectrum. The SLA (Fig. 4b) and sucrose content (Fig. 4c) models both had influential peaks in
172
the visible wavelengths that overlapped with the chlorophyll model, but also showed trait-
173
specific contributions throughout the near and short-wavelength infrared regions. The
174
chlorophyll and Vmax models were very similar in the visible portion of the spectrum, with
175
moderate differences between 800-1750 nm (Fig. 4d).
176
177
As an alternative to PLSR model building, chlorophyll, N and SLA were also estimated
using previously published simple indices and correlated with direct measures of traits (Table 1).
7
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
178
Several indices, including mNDVI, mSRCHL, SIPI and SR2, were unable to predict chlorophyll
179
content accurately in maize leaves (r2 = 0.064, 0.006, 0.067, 0.172). Estimated chlorophyll
180
values from other indices were correlated with measured values (Table 1), but even the best of
181
these indices, based on the coefficient of determination (Datt4, r2 = 0.74), was outperformed by
182
the PLSR model (r2 = 0.85, Fig. 1b). Trait values estimated using previously published indices
183
for N content and SLA showed poor correlations with measured values and were outperformed
184
by PLSR analysis of reflectance spectra (Table 1, Fig. 1).
185
186
High-throughput Screening of Diverse Inbred and Hybrid Maize for O3 Response
187
Once it was determined that PLSR models could reasonably predict foliar biochemical
188
and physiological traits, we tested the approach for high-throughput phenotyping potential by
189
screening diverse inbred and hybrid maize lines grown in the field at ambient and elevated [O3]
190
in 2013, 2014 and 2015 (Supplemental Table 4). Using two spectroradiometers, we were able to
191
collect leaf reflectance spectra from maize plants grown in 1024 rows spread across a 16 ha field
192
in 2-4 days while limiting the period of measurement to the middle 4 hours of each day (~11:00
193
a.m. – 3:00 p.m.). By contrast, destructive sampling from each of the rows of the experiment
194
required 12 people to sample for 4-5 hours over 2 days, making sample collection approximately
195
6 times more efficient for spectral traits. The time savings of spectral reflectance over wet
196
chemistry were even more dramatic for post-processing of samples and data. Analysis of spectral
197
data took approximately 10 days, while processing and analysis of samples for wet chemistry
198
and leaf gas exchange took between 30-90 days depending on the trait.
199
Diverse inbred and hybrid maize genotypes showed a 1.3 – 2.2 fold range of values for
200
chlorophyll content, N content, SLA, sucrose content and Vmax, as estimated by the PLSR models
201
(Fig. 5). For example, in 2014 chlorophyll content ranged from 15.2-31.2 µg cm-2 (Fig. 5a), N
202
content from 3.2-4.3 % (Fig. 5b), SLA from 20.6-29.6 mm2 mg-1 (Fig. 5c), sucrose content from
203
3.8-5.7 µmol cm-2 (Fig. 5d), and Vmax from 18.8-41.1 µmol m-2 s-1 (Fig. 5e) in the 52 inbred
204
genotypes grown at ambient [O3]. In general, the range of trait values for hybrids grown in
205
ambient [O3] was narrower than for inbreds grown at ambient [O3] (Fig. 5), as may be expected
206
because all of the hybrids shared a common parent in B73 (Supplemental Table 4). Hybrid
207
genotypes also had greater average chlorophyll, N and Vmax compared to inbred genotypes, and
208
lower average sucrose content (Fig. 5d). Inbred and hybrid genotypes grown at elevated [O3]
8
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
209
showed significant reductions in chlorophyll content in all years (Fig. 5a; Table II), and while
210
there was some inter-annual variation, elevated [O3] also decreased N content and Vmax (Fig. 5b,
211
e; Table II), and increased SLA (Fig. 5c; Table II).
212
213
214
DISCUSSION
This study demonstrates the ability to rapidly and cost-effectively screen for intraspecific
215
variation in maize leaf biochemical and physiological traits using PLSR models applied to leaf
216
reflectance spectra. Leaf chlorophyll content, N content, SLA, sucrose content and Vmax could be
217
predicted from reflectance spectra of maize (Figs. 1, 2). This capability to rapidly and
218
simultaneously capture five of the most commonly assessed and important leaf traits to plant
219
carbon and nitrogen relations in a single spectrum represents a significant advance with a wide
220
range of potential applications in maize genetics and breeding. Previous attempts to use remote
221
sensing to investigate O3 damage in maize measured visual damage to leaves and canopies (Kraft
222
et al., 1996; Rudorff et al., 1996, Runeckles and Resh, 1975; see Meroni et al., 2009 for review),
223
most often in terms of the normalized difference vegetation index, which is based on differences
224
in surface reflectance in the near infrared and visible ranges of the spectrum. This was also
225
limited to canopy-level remote sensing of a single maize genotype. The current study
226
demonstrated success in early detection and evaluation of leaf-level biochemical and
227
physiological responses when visual symptoms of O3 stress were minimal. The proof-of-concept
228
for the use of spectral reflectance to detect phenotypic variation within a C4 species is a
229
significant advance over previous uses of similar techniques which have distinguished larger
230
species-level differences in natural or agricultural ecosystems (Asner and Martin, 2008; Asner et
231
al., 2011; Serbin et al., 2012, 2014, 2015).
232
233
234
High-throughput Phenotyping with Hyperspectral Reflectance
High-throughput field phenotyping efforts have successfully investigated morphological,
235
developmental and agronomic traits in large numbers of crop genotypes. However, rapid
236
measurements of physiological traits are needed to fully understand plant-environment
237
interactions, including plant stress response (Großkinsky et al., 2015). For field phenotyping to
238
be maximally useful, approaches should be high-throughput, precise and multi-dimensional,
239
providing accurate information about relevant traits (Dhondt et al., 2013). The hyperspectral
9
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
240
reflectance and PLSR modeling approach taken in this study meets those goals. Collection of
241
leaf reflectance spectra takes less than a minute per leaf, and is non-destructive so it is possible
242
monitor a leaf over its lifetime. For estimating Vmax from gas exchange, it could take at least 20-
243
30 minutes to assess the response of A to variation in intercellular [CO2] (ci) curve on a single
244
leaf. Even measurements of maximum light and CO2-saturated photosynthesis (Amax) could take
245
5 – 10 minutes per leaf in the field where the leaf has to acclimate to the cuvette conditions,
246
which may differ from atmospheric conditions. Generating parameter estimates by fitting a
247
photosynthesis model to the measured gas exchange data would take approximately the same
248
amount of time as fitting the PLSR model. Therefore, we estimate that the reflectance approach
249
for estimating Vmax is at least 10 times faster in terms of data collection and can be completed
250
during a shorter period of time during the day to avoid circadian or diurnal effects on leaf traits
251
that might confound measurements taken over a longer period of time (Dodd et al., 2005). For
252
other traits like sucrose content or N content, “gold standard” measurements require a destructive
253
leaf sample for biochemical analysis, which is lower throughput and does not allow repeated
254
monitoring over time. Estimation of numerous traits from a single spectrum provides multi-
255
dimensionality, and the speed with which spectra can be collected mean that at least temporally,
256
measurements can be taken with high resolution. In this study, we derived leaf-level biochemical
257
and structural traits from leaf reflectance spectra collected with a leaf clip that provided a
258
uniform light environment, thus minimizing some of the environmental and spatial variation that
259
canopy level reflectance measurements inevitably have. The leaf clip facilitates investigation of
260
tissue-level phenomena and the effects of leaf age, but there is a trade-off in that further study
261
would be needed to scale these leaf-level parameters to a canopy.
262
263
264
Leaf-level Predictive Models
Leaf-level hyperspectral data have been used previously to estimate photosynthetic
265
parameters in C3 species (Doughty et al. 2011; Serbin et al. 2012), but not in C4 species which
266
have a carbon concentrating mechanism and different biochemical limitations to photosynthetic
267
rate than C3 species. For example, while light-saturated photosynthetic rate is limited by
268
carboxylation capacity at low intercellular [CO2] in C3 plants (Farquhar et al., 1980), it is instead
269
limited by PEPcase activity at low intercellular [CO2] in C4 plants (von Caemmerer, 2000). C3
270
and C4 species also invest differentially in photosynthetic proteins and pigments, and have leaves
10
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
271
with distinct morphological and structural characteristics. Based on these differences, a number
272
of studies have used remote sensing approaches to distinguish C3 and C4 species experimentally
273
or on the landscape (Irisarri et al., 2009; Liu and Cheng, 2011; Adjorlolo et al., 2012). Perhaps it
274
is not surprising that a new PLSR model was needed to estimate parameters related to C4
275
photosynthesis in maize, especially given that predictions of photosynthetic capacity are not
276
direct expressions of the reflectance spectra, but instead are due to a combination of anatomical
277
(e.g., specific leaf area) and chemical factors (e.g., nitrogen or chlorophyll) that are more directly
278
expressed in reflectance spectra (Jacquemoud and Baret, 1990). In our study, we could not use
279
reflectance spectra to accurately predict Vp,max in maize (Fig. 1i, j), but could accurately predict
280
Vmax (Fig. 1g, h). It will be interesting to test if this is also the case for other C4 species.
281
Similar to previous PLSR models for C3 photosynthetic parameters (Serbin et al., 2011),
282
the C4 PLSR model for Vmax used wavebands in the visible region of the spectrum where
283
chlorophyll and other pigments have strong features, but also used wavebands in the near
284
infrared and short-wave infrared regions (Fig. 4), indicating that Vmax is not just simply a
285
function of chlorophyll content. We also observed along with previous studies (Hansen &
286
Schjoerring, 2003; Doughty et al., 2011; Serbin et al., 2011) that both the measured and PLSR-
287
modeled leaf traits were significantly correlated (Fig. 3). Relationships among leaf structural and
288
biochemical traits also converge across a very diverse range of species, and photosynthesis is
289
positively correlated to leaf N content and SLA (Reich et al., 1997; Wright et al., 2004). Global
290
trait databases also have reported correlations between leaf N content and specific leaf area
291
(Reich et al., 1997); however, this correlation was not significant within diverse maize lines (Fig.
292
3). This may be because within-species variation is more constrained than across-species
293
variation in leaf traits. But, most importantly, the same conclusions about correlation among
294
traits and variation in leaf traits due to genotype or treatment were drawn from tests of measured
295
and modeled data (Fig. 3; Supplemental Fig. 1, Supplemental Table 3).
296
PLSR models developed for maize also improved predictions of chlorophyll content, N
297
content and SLA compared to previously published simple indices (Table 1). The performance
298
improvement of the PLSR model compared to simple indices has been previously reported
299
(Hansen and Schjoerring, 2003; Atzberger et al., 2010) and has several contributing sources.
300
First, PLSR takes advantage of the full spectral information, most of which is excluded by
301
simple indices, yet clearly contributes to improved trait prediction based on underlying
11
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
302
biophysical or biochemical attributes (Hansen and Schjoerring, 2003). PLSR is also relatively
303
insensitive to sensor noise (Atzberger et al., 2010). The other advantage is building models
304
specifically to the species under study, in this case maize, while the other simple indices were
305
built using other species or diverse species (e.g., Datt, 1988; Vogelmann et al., 1993; Maccioni et
306
al., 2001; le Maire et al., 2004, 2008). Of course, this may be a disadvantage as well, and further
307
studies need to test whether the PLSR models developed for maize in this study can also be
308
accurately used in other C4 species.
309
310
311
High-throughput Phenotyping of Stress Response in Maize
This study used diverse maize lines grown at elevated [O3] as a proof of concept testbed
312
for high-throughput phenotyping of leaf biochemical traits using leaf reflectance spectra. The
313
maize inbred lines represent much of the genetic diversity within the species (McMullen et al.,
314
2009), and elevated [O3], as a treatment, provides an attractive test for this approach environment
315
because common phenotypes associated with O3 damage include reduced photosynthetic
316
capacity, leaf N content, chlorophyll content, increased respiration rates and levels of
317
antioxidants and altered specific leaf area (Reich & Amundson, 1985; Fiscus et al., 2005;
318
Betzelberger et al., 2012). Based on the biochemical and physiological traits estimated from
319
hyperspectral reflectance, both maize inbreds and hybrids were sensitive to elevated [O3] and
320
showed reduced chlorophyll, N content and photosynthetic capacity (Fig. 5; Table II). This was
321
accompanied by greater SLA, indicating reduced leaf mass per unit area, which is often
322
associated with lower photosynthetic capacity and/or lower starch content (Poorter et al., 2012).
323
This is consistent with accelerated leaf senescence at elevated [O3] observed in other species
324
(Miller et al., 1999; Ainsworth et al., 2012; Gillespie et al., 2012) and with reduced carbon gain
325
being at least partly responsible for the 10% reduction in yield of U.S. maize due to O3 over the
326
last 30 years (McGrath et al., 2015). This study also showed that using hyperspectral reflectance,
327
it was possible to detect genotypic differences in physiological and biochemical leaf traits across
328
maize lines, and variation in their response to elevated [O3] (Fig. 6). The ability to rapidly detect
329
this variation in the field across a diverse panel of genotypes is a key first step to improving
330
abiotic stress tolerance in maize.
331
332
CONCLUSIONS
12
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
333
Rapid measurement of physiological traits is widely anticipated as a transformative technology
334
for the experimental research needed to understand mechanisms of crop stress response and to
335
maximize the advantage of genetic resources for crop improvement. Here we provide evidence
336
that leaf-level physiological and biochemical traits that are hallmark markers of stress response
337
can be accurately predicted from PLSR models built from leaf reflectance spectra in maize,
338
thereby providing a high-throughput, non-destructive method for phenotyping.
339
13
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
340
MATERIALS AND METHODS
341
Greenhouse Experiment
342
Maize (Zea mays L.) genoytpe B73 was planted on 15 March 2014 in 17 cm dia. pots
343
containing potting mix (Sunshine LC1 mix; Sun Gro Horticulture Distribution Inc., Bellevue,
344
WA). Plants were grown in the greenhouse and fertilized with 40% Long Ashton nutrient
345
solution (Hewitt & Smith, 1975) with either 6 mM NH4NO3 (Ample N) or 0.25 mM NH4NO3
346
(Low N) starting 20 days after planting (DAP). The greenhouse was sunlit and additional
347
supplemental lighting from metal halide lamps provided a minimum PPFD of 200 µmol m-2 s-1
348
from 06:00-20:00. Foliar tissue and hyperspectral measurements were collected 55-60 DAP from
349
a developmental range of leaves at V7-V10.
350
351
352
Field Experiments
Diverse inbred and hybrid maize genotypes (Supplemental Table 4) were grown in
353
ambient and elevated [O3] at the Free Air Concentration Enrichment field site in Champaign, IL
354
(www.igb.illinois.edu/soyface/) during the growing seasons of 2013, 2014 and 2015. Maize
355
genotypes were planted with a precision planter in 3.35 m rows at a density of 8 plants m-1 and
356
row spacing of 0.76 m. In 2013, 203 inbred genotypes were grown at ambient and elevated [O3]
357
(n=2) in octagonal ‘rings’ ~20 m in diameter. Each individual genotype was planted in a single
358
row plot within a ring. Placement of individual genotypes was randomized between the two
359
replicates, and the location of each genotype was fixed within a pair of ambient and elevated
360
[O3] rings (‘ring pair’) that were located near one another in the field (but still separated by ~100
361
m). B73 was planted in 12 locations within each ring in 2013. In 2014, 52 inbred and 26 hybrid
362
genotypes were grown at ambient and elevated [O3] (n=4). Each inbred genotype was planted in
363
a single row within each ring, and location of genotypes was randomized among the four
364
replicates, but conserved within an ambient and elevated [O3] ‘ring pair.’ Each hybrid genotype
365
was planted in two rows within each ring and randomized as described above for inbred
366
genotypes. In 2014, B73 was planted in 10 rows within each inbred ring and B73 x Mo17 was
367
planted in 14 rows within each hybrid ring. In 2015, 10 inbred and 8 hybrid genotypes were
368
grown at ambient and elevated [O3] (n=4). Each genotype was planted in 5 rows within each
369
octagonal ring with a spatial design ensuring that each genotype was planted in one row in the
370
central, north, south, east and west side of the ring.
14
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
371
Ozone was increased to a target set-point of 100 nL L-1 from ~10:00-18:00 throughout
372
the growing seasons using free air O3 enrichment technology, as described in Morgan et al.
373
(2004) with the following modifications. Ozone was produced by passing high voltage through
374
O2 in O3 generators (CFS-3 2G, Ozonia, Leonia, NJ, USA), then added to compressed air for
375
fumigation. Ozone-enriched air was released on the upwind sides of the octagon from horizontal
376
pipes surrounding the rings. Within each ring, there was a 1m buffer area which was not used for
377
experiments. Fumigation was stopped when leaves were wet (during rain) and when wind speed
378
was too low (< 0.5 m s-1) to maintain consistent treatment. The average daily [O3] and
379
cumulative [O3] exposure during the 2013, 2014 and 2015 growing seasons are summarized in
380
Supplemental Table 5.
381
382
383
Training Data Set
Training data sets were collected to pair leaf reflectance spectra with “gold standard”
384
biochemical and physiological measurements of chlorophyll content, N content, SLA, Vmax,
385
Vp,max, sucrose content and ORAC. Measurements for the training data set were collected from
386
diverse maize lines grown at ambient and elevated [O3] in the field in 2014 and 2015 and from
387
inbred line B73 grown in the greenhouse at high and low N availability. The training data sets for
388
each of the successful PLSR traits are provided in Supplemental Dataset 1.
389
390
391
“Gold Standard” Methods for Measuring Leaf Physiology and Biochemistry
To determine Vp,max and Vmax for the training data set, the response of net assimilation (A)
392
to intercellular [CO2] (Ci) was measured across a range of [CO2] from 25 to 1,500 µmol mol-1
393
using a portable photosynthesis system (LI-6400; LICOR Biosciences, Lincoln, NE) set at
394
constant PPFD (2,000 µmol m-2 s-1). The ~6 initial points of the A/Ci curve (Ci < 60 µmol mol-1)
395
were fit to a model for PEPC-limited photosynthesis (von Caemmerer, 2000) and used to
396
estimated Vp,max. Vmax was estimated as the horizontal asymptote of a 4-parameter nonrectangular
397
hyperbolic function (Markelz et al., 2011). Before starting each A/Ci curve, leaf temperature was
398
measured with an infrared thermometer (62 MAX; Fluke Corporation, Everett, WA) and used to
399
set the leaf chamber cuvette at ambient conditions. This also ensured that gas exchange
400
measurements were done at the same temperature as leaf reflectance spectra. For the greenhouse
401
experiment, two intact leaves were measured per plant; the first and third youngest leaf with a
15
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
402
visible collar. For field samples, the third youngest leaf with a visible collar was excised and
403
immediately re-cut under water prior to dawn.
404
Immediately after each A/Ci curve was completed, tissue was removed from the leaf
405
using a cork borer and flash frozen in liquid nitrogen. One leaf disk (~1.4 cm2) was incubated in
406
96% ethanol (v/v) for 3 days at 4°C in order to determine chlorophyll content using the equations
407
in Lichtenthaler and Wellburn (1983). Another leaf disk was used to determine the oxygen
408
radical absorbance capacity (ORAC) content following the procedures in Gillespie et al. (2007).
409
A third leaf disk was used to determine sucrose content according to the methods of Jones et al.
410
(1977). Specific leaf area (SLA) was determined by weighing a known area of leaf tissue after
411
drying for five days at 60°C. The dried leaf tissue was then ground to a fine powder and
412
combusted with oxygen in an elemental analyzer (Costech 4010, Costech Analytical
413
Technologies, Inc., Valencia, CA). N content was determined by comparing experimental
414
samples to an acetanalide standard curve.
415
416
417
Maize Leaf Reflectance Spectroscopy
Before destructively sampling leaf tissue, reflectance spectra were captured from the
418
adaxial surface of the same leaves that were used for “gold-standard” measures of leaf
419
physiology and biochemistry using a full-range spectroradiometer (ASD FieldSpec 4 Standard
420
Res, Analytical Spectral Devices, Boulder, CO) equipped with an illuminated leaf clip contact
421
probe. The scanning time for the instrument is ~100 ms, and each spectrum collected was the
422
average of 10 scans. A reflective white reference (Spectralon, Labsphere Inc., North Dutton, NH)
423
was used to standardize relative reflectance of the samples. The spectroradiometer was turned on
424
for at least 30-60 minutes before reflectance measurements began to allow for the three arrays to
425
warm up. Reflectance measurements were obtained from the same location on the leaf as the
426
A/Ci curve measurements and were restricted to the central section of the leaf avoiding the
427
midrib. A preliminary test calculated bootstrapped (100 fold) mean and variance values from a
428
population of 25 measurements collected from a single leaf, from which we determined that six
429
measurements per leaf reduced sample variance by over 50% (Supplemental Fig. 2).
430
Subsequently, all leaf reflectance measurements used the average of six spectra from each leaf
431
taken at different locations within a ~7 cm band in the central section of the leaf. Before
432
averaging the six spectra collected from a single leaf, a splice correction was applied to the
16
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
433
spectral observations to ensure continuous data across detectors. Data were interpolated to
434
provide 1-nm bandwidths and quality control was applied using the FieldSpectra package in R
435
(Serbin et al., 2014).
436
437
Partial Least-Squares Regression Models
438
Partial least-squares regression (PLSR) was used to develop predictive models with the
439
open-source PLS package (Mevik and Wehrens, 2007) in R following the methods of Serbin et
440
al. (2014). This approach identifies latent factors, or components, that account for most of the
441
variation in a trait variable and generates a linear model consisting of waveband scaling
442
coefficients to transform full-spectrum data. To maximize the predictive ability of the model
443
while minimizing the possibility of overfitting, the predicted residual sum of squares (PRESS)
444
statistic was used to determine the optimal number of model components (Chen et al., 2004). The
445
number of components that minimized the root mean square error (RMSE) of the PRESS statistic
446
was chosen (Wold et al., 2001). We calculated the PRESS statistic through a leave-one-out
447
cross-validation approach, which trains the model on all but one observation, then makes a
448
prediction for the left-out observation. The average error is then computed and used to evaluate
449
the model (Siegman and Jarmer, 2015). Variable importance of the projection (VIP) scores
450
(Wold, 1994) were also determined to compare the relative significance of each wavelength in its
451
contribution to the final model. The spectrum range for all models was 500-2400 nm, except N
452
content, which was restricted to 1500-2400 nm because of well-known attributes of nitrogen
453
absorption (Curran, 1989; Serbin et al., 2014). Performance parameters were generated to assess
454
the predictive ability of each model, including the coefficient of determination (R2) from the
455
leave-one-out cross-validation method, RMSE and model bias.
456
A second evaluation of the PLSR model was performed with a holdout dataset. In this
457
case 80% of the data were used as a training dataset for the estimation of leaf traits, and 20% of
458
the data were used as a testing dataset.
459
For chlorophyll, N content, SLA, Vmax and Vp,max training data sets were collected from
460
both field-grown and greenhouse-grown maize plants. For these traits, two PLSR models were
461
built, one with field-grown data only and a second with the entire training data set from the field
462
and the greenhouse (Fig. 1). The better of the two PLSR models based on larger R2 and smaller
463
RMSE (typically the one with data from both the field and greenhouse) was then applied to the
17
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
464
O3 screening data set collected in the field in 2013, 2014 and 2015. The PLSR model coefficients
465
for each trait are provided in Supplemental Table 1.
466
467
468
High Throughput Screening for O3 Response
As a test of the potential to use leaf reflectance spectra for high throughput screening,
469
diverse maize germplasm were measured at the FACE experiment described above in 2013, 2014
470
and 2015. Leaf reflectance spectra were taken on the third youngest leaf with a collar from all
471
experimental rows (1024 in 2013, 768 in 2014, 912 in 2015). In 2013 and 2014, two plants per
472
genotype (row) were sampled within each ambient and elevated [O3] ring, and in 2015, a single
473
plant per row was sampled because each genotype was planted in 5 rows within each ring. Leaf
474
reflectance was measured twice in 2013, from July 14-17 and July 29-30, and from July 14-18,
475
2014 and July 15-16, 2015.
476
477
478
Statistical Analysis of Measured and Predicted Trait Values
In addition to PLSR models, published indices for estimating chlorophyll content, N
479
content and SLA were tested using the training population (Table 1). Correlations between
480
published indices and measured values were tested. Pairwise correlations among the “gold-
481
standard” measured traits and the spectral-derived traits were tested using the training data set.
482
The ability to distinguish statistical differences in measured and PLSR-modeled traits was also
483
tested with a subset of the training dataset. Measured and PLSR-modeled estimates of
484
chlorophyll content, N content, Vmax and SLA from genotypes B73, Mo17 and B73 x Mo17 were
485
used to test for the effects of genotype, O3 and their interaction using a general linear model.
486
The field designs in 2013, 2014 and 2015 differed in the number of genotypes examined
487
and the layout of the FACE and control rings among years. Models were fit for each year
488
separately. For the 2013 data, the linear model Yijkl= μ+t j+ρij+gk+tgij+εijkl was fit where j is the
489
jth treatment (ambient, elevated [O3]), i is the random effect of the ith ringpair (i=1,..4), and k is
490
the kth genotype. The variance of the ring pairs was assumed to be common for all ambient rings
491
and all elevated [O3] rings but not assumed to be the same between them, and an additional
492
covariance parameter for the southwest corner of the rings was included. In 2014, Yijkl=μ+tj+ρij+
493
gk+tgij+εijkl was fit where j is the jth treatment, i is the random effect of the ith ringpair (i=1,..4),
494
and k is the kth genotype. The increased number of replicates in 2014 allowed us to fit the
18
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
495
residuals for the difference in variance structure as ε∼Ν(0,Σ), where Σ is the variance covariance
496
matrix with diagonal elements σj and all off diagonal terms are zero except for a single
497
covariance p to account for spatial variation among observations in the southwest quadrant of the
498
elevated [O3] rings. In 2015, the ring was divided into 5 sectors, and in each sector all of the
499
genotypes were planted. The model Yijkl=μ+tj+ρij+γs(pij)+ bl+gk+tgij+εijkl was fit where Yijkl is the
500
trait value for the ith ringpair (i=1,..4), the sth sector (s=1,…5) nested inside the ringpair for the jth
501
treatment, the lth sector (1 for the sector in the southwest corner, 0 otherwise), the kth genotype.
502
Ringpair and sector are treated as random with all other effects fixed.
503
504
SUPPLEMENTAL DATA
505
Supplemental Table 1. Partial least square regression (PLSR) model coefficients for predicting
506
foliar traits using leaf reflectance spectra from maize.
507
Supplemental Table 2. Performance comparison between predictive models built with either the entire
508
training population (full model) or 80% of the training population (80% model).
509
Supplemental Table 3. Analysis of variance (ANOVA) degrees of freedom (df), F value and
510
significance level (p) for measured and partial least squares regression (PLSR) modeled leaf traits.
511
512
513
514
515
516
Supplemental Table 4. List of inbred and hybrid maize genotypes grown at ambient (~40 ppb)
and elevated [O3] (~100 ppb) in the 2013, 2014 and 2015 growing seasons.
Supplemental Table 5. Seasonal average daily [O3] exposure at the SoyFACE facility in 2013,
2014 and 2015.
517
genotypes B73, B73 x Mo17 and Mo17.
518
Supplemental Figure 2. Bootstrap estimate of mean chlorophyll content and standard deviation from 1 to
519
24 spectra taken on an individual leaf.
520
Supplemental Dataset 1. Raw spectral data for wavelengths used to build PLSR models, along
521
with measured trait value and PLSR predicted trait value.
522
Supplemental Dataset 2. Raw spectral data for the screening dataset collected on diverse maize
523
inbred and hybrid lines grown in the field in ambient and elevated [O3] in 2013, 2014 and 2015.
Supplemental Figure 1. Mean measured and PLSR-modeled leaf traits +/- 1 standard deviation for maize
524
525
ACKNOWLEDGEMENTS
526
We thank Brad Dalsing, Chad Lantz, Kannan Puthuval and Mac Singer for operating the
527
SoyFACE facility. We thank Matt Lyons, Dan Jaskowiak, John Regan, Anna Molineaux, Jessica
19
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
528
Wedow, Taylor Pederson, Chris Moller, Charles Burroughs and Ben Thompson for assistance
529
with field sampling and laboratory measurements.
20
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
530
TABLES
531
532
533
Table I. Estimates of chlorophyll content, nitrogen content and specific leaf area (SLA) calculated using published spectral indices
and compared to measured values. Each index was applied to the same spectra used for the PLSR model training population. For each
index, the predictive formula is shown along with the correlation coefficient, root-mean-square error and reference for each index.
Trait
Chlorophyll
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
Nitrogen
Index
Datt4
Vogelmann2
Maccioni
Double Difference
Vogelmann1
mSR705
mNDVI705
SR3
SR4
SR1
Gitelson
SR2
SIPI
mNDVI
mSRCHL
NRI1510
MCARI1510
NDNI
SLA
mND
NDMI
NDLMA
mSRSLA
SR
ND
Formulation
R672/(R550*R708)
(R734-R747)-(R715+R726)
(R780-R710)/(R780-R680)
(R749-R720)-(R701-R672)
R740/R720
(R750-R445)/(R705-R445)
(R750-R705)/(R750+R705-2R445)
R750/R550
R700/R670
R750/R700
1/R700
R752/R690
(R800-R445)/(R800-R680)
(R800-R680)/(R800+R680-2R445)
(R800-R445)/(R680-R445)
Adj. R2
0.740
0.736
0.714
0.702
0.701
0.680
0.653
0.580
0.498
0.457
0.387
0.172
0.067
0.064
0.006
RMSE
3.23
3.26
3.39
3.46
3.47
3.58
3.73
4.11
4.49
4.67
4.96
5.77
6.12
6.13
6.36
(R1510-R660)/(R1510+R660)
[(R700-R1510)-0.2(R700-R550)]*(R700/R1510)
[log10(1/R1510)log10(1/R1680)]/[log10(1/R1510)+log10(1/R1680)]
0.450
0.239
0.90
1.05
Herrmann et al. (2010)
Herrmann et al. (2010)
0.198
1.08
Serrano et al. (2002)
(R2285-R1335)/[R2285+R1335-(2*R2400)]
(R1649-R1722)/(R1649+R1722)
(R1368-R1722)/(R1368+R1722)
(R2265-R2400)/(R1620-R2400)
R2295/R1500
(R1710-R1340)/(R1710+R1340)
0.177
0.133
0.122
0.073
0.070
0.061
2.43
2.49
2.51
2.58
2.58
2.60
le Marie et al. (2008)
Cheng et al. (2014)
Cheng et al. (2014)
le Marie et al. (2008)
le Marie et al. (2008)
le Marie et al. (2008)
Reference
Datt (1998)
Vogelman et al. (1993)
Maccioni et al. (2001)
le Maire et al. (2004)
Vogelman et al. (1993)
Sims and Gamon (2002)
Sims and Gamon (2002)
Gitelson and Merzlyak (1997)
McMurtey et al. (1994)
Gitelson and Merzlyak (1997)
Gitelson et al. (1999)
Gitelson and Merzlyak (1997)
Penuelas et al. (1995)
Sims and Gamon (2002)
Sims and Gamon (2002)
534
535
21
536
537
538
539
540
541
542
543
Table II. Statistical analysis (p value) of the effects of elevated ozone concentration ([O3]),
genotype (gen) and the interaction of [O3] and genotype on leaf physiological and biochemical
traits estimated from leaf reflectance spectra. Box plots of leaf traits are shown in Figure 5.
Spectra were collected from July 14-17, 2013 (2013a), July 29-30, 2013 (2013b), July 14-18,
2014 (2014) and July 15-16, 2015 (2015). Inbred and hybrid lines were evaluated in different
statistical models. Chl: chlorophyll content, N: leaf N content, Vmax: [CO2]-saturated rate of
photosynthesis, SLA: specific leaf area, Suc: leaf sucrose content
Inbred Lines
2013a
[O3]
Gen
[O3] x Gen
2013b
[O3]
Gen
[O3] x Gen
2014
[O3]
Gen
[O3] x Gen
2015
[O3]
Gen
[O3] x Gen
Hybrid Lines
2014
[O3]
Gen
[O3] x Gen
2015
[O3]
Gen
[O3] x Gen
Chl
N
Vmax
SLA
Suc
<0.0001
<0.0001
0.19
0.17
<0.0001
0.60
0.56
<0.0001
0.05
<0.0001
<0.0001
0.94
<0.0001
<0.0001
0.12
<0.0001
<0.0001
0.90
<0.0001
<0.0001
0.35
0.95
<0.0001
0.72
<0.0001
<0.0001
0.98
0.05
<0.0001
0.99
<0.0001
<0.0001
0.0002
<0.0001
<0.0001
0.05
<0.0001
<0.0001
0.03
<0.0001
<0.0001
<0.0001
0.004
<0.0001
<0.0001
<0.0001
<0.0001
0.17
<0.0001
<0.0001
0.32
0.0007
<0.0001
0.42
0.01
<0.0001
0.05
<0.0001
<0.0001
0.25
<0.0001
0.0005
0.07
<0.0001
0.12
0.40
<0.0001
0.0008
0.22
0.08
<0.0001
0.86
<0.0001
<0.0001
0.29
<0.0001
<0.0001
0.74
0.68
<0.0001
0.90
<0.0001
<0.0001
0.33
0.0007
0.004
0.74
<0.0001
0.02
0.75
544
545
546
22
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
547
FIGURE LEGENDS
548
549
Figure 1. Measured vs. partial least-squares regression (PLSR) modeled values of (a-b) leaf
550
chlorophyll content, (c-d) N content, (e-f) specific leaf area, (g-h) [CO2]-saturated rate of
551
photosynthesis (Vmax), and (i-j) maximum rate of PEP carboxylation (Vp,max). PLSR models were
552
built using a training population consisting of (a, c, e, g, i) genotypically diverse field-grown
553
plants exposed to ambient and elevated [O3], or (b, d, f, h, j) a combination of the field-grown
554
plants and greenhouse plants (genotype B73) grown in ample or limiting N. In each panel the
555
cross-validation R2, root mean square error and bias of the model is shown along with the size of
556
the validation population and number of model components (comp) used in each PLSR model.
557
The gray dashed line shows the 1:1 line.
558
559
Figure 2. Measured vs. partial least-squares regression (PLSR) modeled values of (a) leaf
560
sucrose content and (b) leaf total oxygen radical absorbance capacity (ORAC). PLSR models
561
were built using a training population consisting of diverse field-grown maize inbred and hybrid
562
lines exposed to ambient and elevated [O3]. In each panel the cross-validation R2, root mean
563
square error and bias of the model is shown along with the size of the validation population and
564
number of model components (comp) used in each PLSR model. The gray dashed line shows the
565
1:1 line.
566
567
Figure 3. Pairwise correlations of measured (black symbols) and modeled (gray symbols) leaf
568
traits. Modeled traits are partial least squares predictions from leaf reflectance spectra collected
569
from diverse maize inbred and hybrid lines.
570
571
Figure 4. The variable importance of the projection (VIP) across the spectral region for each of
572
the PLSR models. VIP scores for the chlorophyll model are compared to models for (a) N
573
content (%), (b) specific leaf area (SLA), (c) sucrose and (d) Vmax. A VIP score > 0.8 is
574
considered to influence the trait estimate (Wold, 1994). The orange shaded area in panel (a)
575
indicates that the model for N content was built using a truncated spectra region (1500-2400 nm).
576
All other models utilized the spectrum from 500-2400 nm.
577
23
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
578
Figure 5. Box plots showing the distribution of values estimated from PLSR models for diverse
579
maize genotypes including (a) foliar chlorophyll content, (b) N content, (c) specific leaf area
580
(SLA), (d) sucrose content and (e) photosynthetic capacity (Vmax). Maize inbred (IN) and hybrid
581
(HY) genotypes were grown at ambient (Amb, ~40 ppb) and elevated [O3] (Elev, 100 ppb) in
582
each of the three growing seasons. Spectra were collected from July 14-17, 2013 (2013a), July
583
29-30, 2013 (2013b), July 14-18, 2014 and July 15-16, 2015. The number of individual rows
584
successfully sampled during each time point is shown in parentheses. Asterisks indicate
585
significant effects of elevated [O3] within a given time point. **p<0.01, ***p<0.001
586
587
588
589
24
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
590
LITERATURE CITED
591
Adjorlolo C, Cho MA, Mutanga O, Ismail R (2012) Optimizing spectral resolutions for the
592
classification of C3 and C4 grass species, using wavelengths of known absorption features. J
593
Appl. Remote Sens. 6: 063560.
594
Ainsworth EA, Beier C, Calfapietra C, Ceulemans R, Durand-Tardif M, Farquhar GD,
595
Godbold DL, Hendrey GR, Hickler T, Kaduk J, et al. (2008) Next generation of elevated
596
[CO2] experiments with crops: a critical investment for feeding the future world. Plant Cell
597
Environ 31: 1317-1324
598
Ainsworth EA, Serbin SP, Skoneczka JA, Townsend PA (2014) Using leaf optical properties
599
to detect ozone effects on foliar biochemistry. Photosyn Res 119: 65-76.
600
Ainsworth EA, Yendrek CR, Sitch S, Collins WJ, Emberson LD (2012) The effects of
601
tropospheric ozone on net primary productivity and implications for climate change. Ann Rev
602
Plant Biol 63: 637-661
603
Araus JL, Cairns JE (2014) Field high-throughput phenotyping: the new crop breeding frontier.
604
Trends Plant Sci 19: 52-61
605
Asner GP, Martin RE (2008) Spectral and chemical analysis of tropical forests: scaling from
606
leaf to canopy levels. Remote Sens Environ 112: 3958–3970
607
Asner GP, Martin RE (2015) Spectroscopic remote sensing of non-structural carbohydrates in
608
forest canopies. Remote Sensing 7: 3526-3547
609
Asner GP, Martin RE, Knapp DE, Tupayachi R, Anderson C, Carranza L, Martinez P,
610
Houcheime M, Sinca F, Weiss P (2011) Spectroscopy of canopy chemicals in humid tropical
611
forests. Remote Sens Environ 115: 3587–3598
612
Atzberger C, Guerif M, Baret F, Werner W (2010) Comparative analysis of three
613
chemometric techniques for the spectroradiometric assessment of canopy chlorophyll content in
614
winter wheat. Comput Electron Agric 73: 165-173
615
Betzelberger AM, Yendrek CR, Sun J, Leisner CP, Nelson RL, Ort DR, Ainsworth EA
616
(2012) Ozone exposure response for U.S. soybean cultivars: linear reductions in photosynthetic
617
potential, biomass, and yield. Plant Physiol 160: 1827-1839
618
Brosché M, Merilo E, Mayer F, Pechter P, Puzõrjova I, Brader G, Kangasjärvi J, Kollist H
619
(2010) Natural variation in ozone sensitivity among Arabidopsis thaliana accessions and its
620
relation to stomatal conductance. Plant Cell Environ 33: 914-925
25
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
621
Chen S, Hong X, Harris CJ, Sharkey PM (2004) Spare modeling using orthogonal forest
622
regression with PRESS statistic and regularization. IEEE Trans Syst Man Cybern 34: 898-911
623
Cheng T, Rivard B, Sanchez-Azofeifa AG, Feret JB, Jacquemoud S, Ustin SL (2014)
624
Deriving leaf mass per area (LMA) from foliar reflectance across a variety of plant species using
625
continuous wavelet analysis. ISPRS J Photogramm Remote Sens 87: 28-38
626
Christensen JH, Hewitson B, Busuioc A, Chen A, Gao X, et al. (2007) Regional climate
627
projections. In Solomon S, Qin D, Manning M, Chen Z, Marquis M, Averyt KB, Tignor M,
628
Miller HL, eds, Climate Change 2007: The Physical Science Basis. Contribution of Working
629
Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change.
630
Cambridge University Press, Cambridge, UK, and New York, pp 847–940
631
Curran PJ (1989) Remote sensing of foliar chemistry. Remote Sens Environ 30: 271-278
632
Datt B (1998) Remote sensing of chlorophyll a, chlorophyll b, chlorophyll a+b and total
633
carotenoid content in eucalyptus leaves. Remote Sens Environ 66: 111-121
634
Dhondt S, Wuyts N, Inzé D (2013) Cell to whole-plant phenotyping: the best is yet to come.
635
Trends Plant Sci 18: 428-439
636
Dizengremel P, Le Thiec D, Bagard M, Jolivet Y (2008) Ozone risk assessment for plants:
637
central role of metabolism-dependent changes in reducing power. Environ Pollut 156: 11–15
638
Dodd AN, Salathia N, Hall A, Kévei E, Tóth R, Nagy F, Hibberd JM, Millar AJ, Webb
639
AAR (2005) Plant circadian clocks increase photosynthesis, growth, survival, and competitive
640
advantage. Science 309: 630-633
641
Doughty CE, Asner GP, Martin RE (2011) Predicting tropical plant physiology from leaf and
642
canopy spectroscopy. Oecologia 165: 289-299
643
Dreccer MF, Barnes LR, Meder R (2014) Quantitative dynamics of stem water soluble
644
carbohydrates in wheat can be monitored in the field using hyperspectral reflectance. Field Crops
645
Res 159: 70-80
646
Farquhar GD, von Caemmerer S, Berry JA (1980) A biochemical model of photosynthetic
647
assimilation in C3 species. Planta 149: 78–90
648
Fiscus EL, Booker FL, Burkey KO (2005) Crop responses to ozone: uptake, modes of action,
649
carbon assimilation and partitioning. Plant Cell Environ 28: 997-1011
26
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
650
Frei M, Tanaka JP, Wissuwa M (2008) Genotypic variation in tolerance to elevated ozone in
651
rice: dissection of distinct genetic factors linked to tolerance mechanisms. J Exp Bot 59: 3741-
652
3752
653
Furbank RT, Tester M (2011) Phenomics - technologies to relieve the phenotyping bottleneck.
654
Trends Plant Sci 16: 635-644
655
Gamon JA, Serrano L, Surfus JS (1997) The photochemical reflectance index: an optical
656
indicator of photosynthetic radiation use efficiency across species, functional types, and nutrient
657
levels. Oecologia 112: 492–501
658
Gillespie KM, Chae JM, Ainsworth EA (2007) Rapid measurement of total antioxidant
659
capacity in plants. Nature Protocols 2: 867-870
660
Gillespie KM, Xu F, Richter KT, McGrath JM, Markelz RJC, Ort DR, Leakey ADB,
661
Ainsworth EA (2012) Greater antioxidant and respiratory metabolism in field-grown soybean
662
exposed to elevated O3 under both ambient and elevated CO2. Plant Cell Environ 35: 169-184
663
Gitelson AA, Buschmann C, Lichtenthaler HK (1999) The chlorophyll fluorescence ratio
664
F735/F700 as an accurate measure of the chlorophyll content in plants. Remote Sens Environ 69:
665
296-302
666
Gitelson AA, Merzlyak MN (1997) Remote estimation of chlorophyll content in higher plant
667
leaves. Int J Remote Sens 18: 2691-2697
668
Großkinsky DK, Svensgaard J, Christensen S, Roitsch T (2015) Plant phenomics and the
669
need for physiological phenotyping across scales to narrow the genotype-to-phenotype
670
knowledge gap. J Exp Bot 66: 5429-5440
671
Hansen PM, Schjoerring JK (2003) Reflectance measurement of canopy biomass and nitrogen
672
status in wheat crops using normalized difference vegetation indices and partial least squares
673
regression. Remote Sens Environ 86: 542-553
674
Herrmann I, Karnieli A, Bonfil DJ, Cohen Y, Alchanatis V (2010) SWIR-based spectral
675
indices for assessing nitrogen content in potato fields. Int J Remote Sens 31: 5127-5143
676
Hewitt EJ, Smith TA (1975) Plant mineral nutrition. English Universities Press Ltd., London.
677
Irisarri JGN, Oesterheld M, Verón SR, Paruelo JM (2009) Grass species differentiation
678
through canopy hyperspectral reflectance. Int J Remote Sens 22: 5959-5975
679
Jacquemoud S, Baret F (1990) PROSPECT: a model of leaf optimal properties spectra. Remote
680
Sens Environ 34: 75-91
27
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
681
Jones MGK, Outlaw WH, Lowry OH (1977) Enzymic assay of 10-7 to 10-14 moles of sucrose
682
in plant tissues. Plant Physiol 60: 379-383
683
Kraft M, Weigel H-J, Mejer G-J, Brandes F (1996) Reflectance measurements of leaves for
684
detecting visible and non-visible damage to crops. J Plant Physiol 148: 148-154
685
Kim KM, Kwon YS, Lee JJ, Eun MY, Sohn JK (2004) QTL mapping and molecular marker
686
analysis for the resistance of rice to ozone. Molecules Cells 17: 151-155
687
le Maire G, Francois C, Soudani K, Berveiller D, Pontailler JY, Breda N, Genet H, Davi H,
688
Dufrene E (2008) Calibration and validation of hyperspectral indices for the estimation of
689
broadleaved forest leaf chlorophyll content, leaf mass per area, leaf area index and leaf canopy
690
biomass. Remote Sens Environ 112: 3846-3864
691
le Maire G, Francois C, Dufrene E (2004) Towards universal broad leaf chlorophyll indices
692
using PROSPECT simulated database and hyperspectral reflectance measurements. Remote Sens
693
Environ 89: 1-28
694
Leitao L, Bethenod O, Biolley JP (2007) The impact of ozone on juvenile maize (Zea mays L.)
695
plant photosynthesis: effects on vegetative biomass, pigmentation, and carboxylases (PEPc and
696
Rubisco). Plant Biol 9: 478-488
697
Lichtenthaler HK, Wellburn AR (1983) Determination of total carotenoids and chlorophylls a
698
and b of leaf extracts in different solvents. Biochem Soc Trans 11: 591-592
699
Liu L, Cheng Z (2011) Mapping C3 and C4 plant functional types using separated solar-induced
700
chlorophyll fluorescence from hyperspectral data. Int J Remote Sens 32: 9171-9183
701
Maccioni A, Agati G, Mazzinghi P (2001) New vegetation indices for remote measurement of
702
chlorophylls based on leaf directional reflectance spectra. J Photochem Photobiol B 61: 52-61
703
Markelz RJC, Strellner RS, Leakey ADB (2011) Impairment of C4 photosynthesis by drought
704
is exacerbated by limiting nitrogen and ameliorated by elevated [CO2] in maize. J Exp Bot 62:
705
3235-3246
706
Martin ME, Aber JD (1997) High spectral resolution remote sensing of forest canopy lignin,
707
nitrogen and ecosystem processes. Ecol Appl 7: 431-443
708
McGrath JM, Betzelberger AM, Wang S, Shook E, Zhu XG, Long SP, Ainsworth EA
709
(2015) An analysis of ozone damage to historical maize and soybean yields in the United States.
710
Proc Natl Acad Sci USA 112: 14390-14395
28
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
711
McMullen MD, Kresovich S, Villeda HS, Bradbury P, Li HH, et al. (2009) Genetic properties
712
of the maize nested association mapping population. Science 325: 737-740
713
McMurtey JE III, Chapelle EW, Kim MS, Meisinger JJ, Corp LA (1994) Distinguish
714
nitrogen fertilization levels in field corns (Zea mays L.) with actively induced fluorescence and
715
passive reflectance measurements. Remote Sens Environ 47: 36-44
716
Meroni M, Panigada C, Rossini M, Picchi V, Cogliati S, Colombo R (2009) Using optical
717
remote sensing techniques to track the development of ozone-induced stress. Environ Pollut 157:
718
1413-1420
719
Mevik BH, Wehrens R (2007) The pls package: principal component and partial least squares
720
regression in R. J Stat Software 18: 1-24
721
Miller JD, Arteca RN, Pell EJ (1999) Senescence-associated gene expression during ozone-
722
induced leaf senescence in Arabidopsis. Plant Physiol 120: 1015-1024
723
Morgan PB, Bernacchi CJ, Ort DR, Long SP (2004) An in vivo analysis of the effect of
724
season-long open-air elevation of ozone to anticipated 20150 levels on photosynthesis in
725
soybean. Plant Physiol 135: 2348-2357
726
Penuelas J, Baret F, Filella I (1995) Semi-empirical indices to assess carotenoids/chlorophyll a
727
ratio from leaf spectral reflectance. Photosynthetica 31: 221-230
728
Penuelas J, Filella I (1998) Visible and near-infrared reflectance techniques for diagnosing
729
plant physiological status. Trends Plant Sci 3: 151-156
730
Poorter H, Niklas KJ, Reich PB, Oleksyn J, Poot P, Mommer L (2012) Biomass allocation to
731
leaves, stems and roots: meta-analyses of interspecific variation and environmental control. New
732
Phytol 193: 30-50
733
Reich PB, Amundson RG (1985) Ambient levels of ozone reduce net photosynthesis in tree and
734
crop species. Science 230: 566-570
735
Reich PB, Walters MB, Ellsworth DS (1997) From tropics to tundra: global convergence in
736
plant functioning. Proc Natl Acad Sci USA 94: 13730-13734
737
Richardson AD, Reeves JB (2005) Quantitative reflectance spectroscopy as an alternative to
738
traditional wet lab analysis of foliar chemistry: near-infrared and mid-infrared calibrations
739
compared. Can J For Res 35: 1122-1130
29
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
740
Rudorff BFT, Mulchi CL, Daughtry CST, Lee EH (1996) Growth, radiation use efficiency,
741
and canopy reflectance of wheat and corn grown under elevated ozone and carbon dioxide
742
atmospheres. Remote Sens Environ 55: 163-173
743
Runeckles VC, Resh HM (1975) The assessment of chronic ozone injury to leaves by
744
reflectance spectrophotometry. Atmos Environ 9: 447-452
745
Serbin SP, Dillaway DN, Kruger EL, Townsend PA (2012) Leaf optical properties reflect
746
variation in photosynthetic metabolism and its sensitivity to temperature. J Exp Bot 63: 489-502
747
Serbin SP, Singh A, Desai AR, Dubois SG, Jablonski AD, Kingdon CC, Kruger EL,
748
Townsend PA (2015) Remotely estimating photosynthetic capacity, and its response to
749
temperature, in vegetation canopies using imaging spectroscopy. Remote Sens Environ 167: 78-
750
87
751
Serbin, SP, Singh A, McNeil BE, Kingdon CC, Townsend PA (2014) Spectroscopic
752
determination of leaf morphological and biochemical traits for northern temperate and boreal tree
753
species. Ecol Appl 24: 1651-1669.
754
Serrano L, Penuelas J, Ustin SL (2002) Remote sensing of nitrogen and lignin in
755
Mediterranean vegetation from AVIRIS data: decomposing biochemical from structural signals.
756
Remote Sens Environ 81: 355-364
757
Siegmann B, Jarmer T (2015) Comparison of different regression models and validation
758
techniques for the assessment of wheat leaf area index from hyperspectral data. Int J Remote
759
Sens 36: 4519-4534
760
Sims DA, Gamon JA (2002) Relationships between leaf pigment content and spectral
761
reflectance across a wide range of species, leaf structures and developmental stages. Remote
762
Sens Environ 81: 337-354
763
Street NR, Tallis MJ, Tucker J, Brosché M, Kangasjärvi J, Broadmeadow M, Taylor G
764
(2011) The physiological, transcriptional and genetic responses of an ozone-sensitive and an
765
ozone tolerant poplar and selected extremes of their F2 progeny. Environ Pollut 159: 45-54
766
Tilman D, Balzer C, Hill J, Befort BL (2011) Global food demand and the sustainable
767
intensification of agriculture. Proc Natl Acad Sci USA108: 20260-20264
768
Ueda Y, Siddique S, Frei M (2015) A novel gene, OZONE-RESPONSIVE APOPLASTIC
769
PROTEIN1, enhances cell death in ozone stress in rice. Plant Physiol 169: 873-889
30
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
770
Vogelman JE, Rock BN, Moss DM (1993) Red edge spectral measurements from sugar maple
771
leaves. Int J Remote Sens 14: 1563-1575
772
von Caemmerer S (2000) Biochemical models of leaf photosynthesis. CSIRO Publishing,
773
Collingwood, Victoria.
774
Weber VS, Araus JL, Cairns JE, Sanchez C, Melchinger AE, Orsini E (2012) Prediction of
775
grain yield using reflectance spectra of canopy and leaves in maize plants grown under different
776
water regimes. Field Crops Res 128: 82-90.
777
Wold S, Sjostrom M, Eriksson L (2001) PLS-regression: a basic tool of chemometrics.
778
Chemometr Intell Lab 58: 109-130
779
Wold S (1994) PLS for multivariate linear modeling. In van de Waterbeernd H, editor.
780
Chemometric methods in molecular design, methods and principles in medicinal chemistry.
781
Verlag-Chemie, Weinheim, Germany, pp 195-218
782
Wright IJ, Reich PB, Westoby M, Ackerly DD, Baruch Z, Bongers F, Cavender-Bares J,
783
Chapin R, Cornelissen JHC, Diemer M, et al. (2004) The worldwide leaf economics spectrum.
784
Nature 428: 821-827
785
31
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
(b)
30
20
2
2
R =0.8111
RMSE=2.64
Bias=-0.0083
n=176
4 comp
10
0
0
10
20
30
40
R =0.8501
RMSE=3.33
Bias=0.0012
n=268
8 comp
50 0
10
-2
Predicted SLA (mm2 mg-1)
20
30
40
2
30
2
1
1
2
Measured SLA (mm2 mg-1)
20
25
30
3
4
5
6 0
1
(g)
2
3
4
5
6
Measured N content (%)
(h)
60
50
40
30
2
2
R =0.5557
RMSE=6.61
Bias=-0.0319
n=156
8 comp
20
10
0
35
Measured SLA (mm2 mg-1)
120 (i)
2
R =0.9557
RMSE=0.23
Bias=-0.0014
n=203
10 comp
Measured N content (%)
R =0.6758
RMSE=1.43
Bias=0.0004
n=269
12 comp
35
2
R =0.9184
RMSE=0.20
Bias=-0.0018
n=127
10 comp
2
0
25
25
3
0
30
20
4
50
(f)
R =0.7765
RMSE=1.12
Bias=0.0005
n=182
11 comp
(d)
5
Meas. chlorophyll (g cm )
(e)
20
6 (c)
-2
Meas. chlorophyll (g cm )
35
Predicted Vp,max (mol m-2 s-1)
Predicted N content (%)
40
Predicted Vmax (mol m-2 s-1)
Predicted chlorophyll (g cm-2)
(a)
50
0
20
40
60 0
Measured Vmax (mol m-2 s-1)
R =0.6542
RMSE=6.60
Bias=-0.0073
n=214
7 comp
20
40
60
Measured Vmax (mol m-2 s-1)
(h)
(j)
100
Ample N
Low N
Elev.[O3]
80
60
40
2
R =0.2181
RMSE=23.70
Bias=0.2340
n=99
2 comp
20
0
0
20 40 60 80 100 120 0
-2
2
R =0.4332
RMSE=20.64
Bias=0.0013
n=177
2 comp
Amb.[O3]
20 40 60 80 100 120
-1
Measured Vp,max (mol m s ) Measured Vp,max (mol m-2 s-1)
Figure 1. Measured vs. partial least-squares regression (PLSR) modeled values of (a-b) leaf
chlorophyll content, (c-d) N content, (e-f) specific leaf area, (g-h) [CO2]-saturated rate of
photosynthesis (Vmax), and (i-j) maximum rate of PEP carboxylation (Vp,max). PLSR models were
built using a training population consisting of (a, c, e, g, i) genotypically diverse field-grown plants
exposed to ambient and elevated [O3], or (b, d, f, h, j) a combination of the field-grown plants and
greenhouse plants (genotype B73) grown in ample or limiting N. In each panel the cross-validation
R2, root mean square error and bias of the model is shown along with the size of the validation
population and number of model components (comp) used in each PLSR model. The gray dashed
line shows the 1:1 line.
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
Predicted sucrose (mol cm-2)
10
(a)
8
6
4
2
R =0.6221
RMSE=0.95
Bias=-0.0210
n=61
5 comp
2
0
0
2
4
6
8
10
Predicted ORAC (nmol cm -2)
Measured sucrose (mol cm-2)
4000
(b)
2
R =0.0986
RMSE=538.62
Bias=-1.7256
n=178
5 comp
3500
3000
2500
2000
1500
1000
Ambient
Elevated
500
00
00
40
00
35
00
30
00
25
00
20
00
15
0
10
50
0
0
Measured ORAC (nmol cm-2)
Figure 2. Measured vs. partial least-squares regression (PLSR) modeled values of (a) leaf
sucrose content and (b) leaf total oxygen radical absorbance capacity (ORAC). PLSR models
were built using a training population consisting of diverse field-grown maize inbred and hybrid
lines exposed to ambient and elevated [O3]. In each panel the cross-validation R2, root mean
square error and bias of the model is shown along with the size of the validation population and
number of model components (comp) used in each PLSR model. The gray dashed line shows the
1:1 line.
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
1
2
3
4
5
6
0
20
40
60 0
2
4
6
8
SLA (mm2 mg-1)
16
r = 0.71
p < 0.0001
r = 0.76
p < 0.0001
20
24
r = -0.37
p < 0.0001
28
32
60
50
40
30
20
-2
0
Sucrose (mol cm-2)
Chl (g cm )
Vmax (mol m-2 s-1)
N content (%)
4
4
3
3
2
2
r = -0.59
p < 0.0001
r = 0.79
p < 0.0001
1
0
60
-2
20
40
20
r = 0.56
p < 0.0001
0
0
8
8
6
6
4
4
r = -0.63
p < 0.0001
2
-1
2
-2
-2
Sucrose (mol cm )
40
0
32
SLA (mm mg )
0
r = -0.26
p < 0.001
28
2
Measured traits
-1
r = 0.74
p < 0.0001
60
-2
-1
Vmax (mol m s )
0
Vmax (mol m s )
1
r = 0.68
p < 0.0001
24
20
16
0 10 20 30 40 50 60 1
Chl (g cm )
-2
2
3
4
N content (%)
5
6
0
20
40
60 0
Vmax (mol m s )
-2
-1
2
4
6
8
10
Sucrose (mol cm )
-2
Figure 3. Pairwise correlations of measured (black symbols) and modeled (gray symbols) leaf
traits. Modeled traits are partial least squares predictions from leaf reflectance spectra collected
from diverse maize inbred and hybrid lines.
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
Modeled traits
5
N content (%)
0
6
5
Sucrose (mol cm )
N content (%)
10
6
Figure 4. The variable importance of the projection (VIP) across the spectral region
for each of the PLSR models. VIP scores for the chlorophyll model are compared to
models for (a) N content (%), (b) specific leaf area (SLA), (c) sucrose and (d) Vmax.
A VIP score > 0.8 is considered to influence the trait estimate (Wold, 1994). The
orange shaded area in panel (a) indicates that the model for N content was built
using a truncated spectra region (1500-2400 nm). All other models utilized the
spectrum from 500-2400 nm.
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
Figure 5. Box plots showing the distribution of values estimated from PLSR models for diverse
maize genotypes including (a) foliar chlorophyll content, (b) N content, (c) specific leaf area
(SLA), (d) sucrose content and (e) photosynthetic capacity (Vmax). Maize inbred (IN) and hybrid
(HY) genotypes were grown at ambient (Amb, ~40 ppb) and elevated [O3] (Elev, 100 ppb) in
each of the three growing seasons. Spectra were collected from July 14-17, 2013 (2013a), July
29-30, 2013 (2013b), July 14-18, 2014 and July 15-16, 2015. The number of individual rows
successfully sampled during each time point is shown in parentheses. Asterisks indicate
significant effects of elevated [O3] within a given time point. **p<0.01, ***p<0.001
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
Parsed Citations
Adjorlolo C, Cho MA, Mutanga O, Ismail R (2012) Optimizing spectral resolutions for the classification of C3 and C4 grass species,
using wavelengths of known absorption features. J Appl. Remote Sens. 6: 063560.
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Ainsworth EA, Beier C, Calfapietra C, Ceulemans R, Durand-Tardif M, Farquhar GD, Godbold DL, Hendrey GR, Hickler T, Kaduk J,
et al. (2008) Next generation of elevated [CO2] experiments with crops: a critical investment for feeding the future world. Plant
Cell Environ 31: 1317-1324
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Ainsworth EA, Serbin SP, Skoneczka JA, Townsend PA (2014) Using leaf optical properties to detect ozone effects on foliar
biochemistry. Photosyn Res 119: 65-76.
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Ainsworth EA, Yendrek CR, Sitch S, Collins WJ, Emberson LD (2012) The effects of tropospheric ozone on net primary productivity
and implications for climate change. Ann Rev Plant Biol 63: 637-661
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Araus JL, Cairns JE (2014) Field high-throughput phenotyping: the new crop breeding frontier. Trends Plant Sci 19: 52-61
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Asner GP, Martin RE (2008) Spectral and chemical analysis of tropical forests: scaling from leaf to canopy levels. Remote Sens
Environ 112: 3958-3970
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Asner GP, Martin RE (2015) Spectroscopic remote sensing of non-structural carbohydrates in forest canopies. Remote Sensing 7:
3526-3547
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Asner GP, Martin RE, Knapp DE, Tupayachi R, Anderson C, Carranza L, Martinez P, Houcheime M, Sinca F, Weiss P (2011)
Spectroscopy of canopy chemicals in humid tropical forests. Remote Sens Environ 115: 3587-3598
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Atzberger C, Guerif M, Baret F, Werner W (2010) Comparative analysis of three chemometric techniques for the
spectroradiometric assessment of canopy chlorophyll content in winter wheat. Comput Electron Agric 73: 165-173
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Betzelberger AM, Yendrek CR, Sun J, Leisner CP, Nelson RL, Ort DR, Ainsworth EA (2012) Ozone exposure response for U.S.
soybean cultivars: linear reductions in photosynthetic potential, biomass, and yield. Plant Physiol 160: 1827-1839
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Brosché M, Merilo E, Mayer F, Pechter P, Puzõrjova I, Brader G, Kangasjärvi J, Kollist H (2010) Natural variation in ozone
sensitivity among Arabidopsis thaliana accessions and its relation to stomatal conductance. Plant Cell Environ 33: 914-925
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Chen S, Hong X, Harris CJ, Sharkey PM (2004) Spare modeling using orthogonal forest regression with PRESS statistic and
regularization. IEEE Trans Syst Man Cybern 34: 898-911
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Cheng T, Rivard B, Sanchez-Azofeifa AG, Feret JB, Jacquemoud S, Ustin SL (2014) Deriving leaf mass per area (LMA) from foliar
reflectance across a variety of plant species using continuous wavelet analysis. ISPRS J Photogramm Remote Sens 87: 28-38
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
Christensen JH, Hewitson B, Busuioc A, Chen A, Gao X, et al. (2007) Regional climate projections. In Solomon S, Qin D, Manning
M, Chen Z, Marquis M, Averyt KB, Tignor M, Miller HL, eds, Climate Change 2007: The Physical Science Basis. Contribution of
Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University
Press, Cambridge, UK, and New York, pp 847-940
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Curran PJ (1989) Remote sensing of foliar chemistry. Remote Sens Environ 30: 271-278
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Datt B (1998) Remote sensing of chlorophyll a, chlorophyll b, chlorophyll a+b and total carotenoid content in eucalyptus leaves.
Remote Sens Environ 66: 111-121
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Dhondt S, Wuyts N, Inzé D (2013) Cell to whole-plant phenotyping: the best is yet to come. Trends Plant Sci 18: 428-439
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Dizengremel P, Le Thiec D, Bagard M, Jolivet Y (2008) Ozone risk assessment for plants: central role of metabolism-dependent
changes in reducing power. Environ Pollut 156: 11-15
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Dodd AN, Salathia N, Hall A, Kévei E, Tóth R, Nagy F, Hibberd JM, Millar AJ, Webb AAR (2005) Plant circadian clocks increase
photosynthesis, growth, survival, and competitive advantage. Science 309: 630-633
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Doughty CE, Asner GP, Martin RE (2011) Predicting tropical plant physiology from leaf and canopy spectroscopy. Oecologia 165:
289-299
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Dreccer MF, Barnes LR, Meder R (2014) Quantitative dynamics of stem water soluble carbohydrates in wheat can be monitored in
the field using hyperspectral reflectance. Field Crops Res 159: 70-80
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Farquhar GD, von Caemmerer S, Berry JA (1980) A biochemical model of photosynthetic assimilation in C3 species. Planta 149: 7890
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Fiscus EL, Booker FL, Burkey KO (2005) Crop responses to ozone: uptake, modes of action, carbon assimilation and partitioning.
Plant Cell Environ 28: 997-1011
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Frei M, Tanaka JP, Wissuwa M (2008) Genotypic variation in tolerance to elevated ozone in rice: dissection of distinct genetic
factors linked to tolerance mechanisms. J Exp Bot 59: 3741-3752
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Furbank RT, Tester M (2011) Phenomics - technologies to relieve the phenotyping bottleneck. Trends Plant Sci 16: 635-644
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Gamon JA, Serrano L, Surfus JS (1997) The photochemical reflectance index: an optical indicator of photosynthetic radiation use
efficiency across species, functional types, and nutrient levels. Oecologia 112: 492-501
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Gillespie KM, Chae JM, Ainsworth EA (2007) Rapid measurement of total antioxidant capacity in plants. Nature Protocols 2: 867-870
Pubmed: Author and Title
CrossRef: Author and Title
Downloaded
Google Scholar: Author Only Title Only Author
and Title from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
Gillespie KM, Xu F, Richter KT, McGrath JM, Markelz RJC, Ort DR, Leakey ADB, Ainsworth EA (2012) Greater antioxidant and
respiratory metabolism in field-grown soybean exposed to elevated O3 under both ambient and elevated CO2. Plant Cell Environ
35: 169-184
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Gitelson AA, Buschmann C, Lichtenthaler HK (1999) The chlorophyll fluorescence ratio F735/F700 as an accurate measure of the
chlorophyll content in plants. Remote Sens Environ 69: 296-302
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Gitelson AA, Merzlyak MN (1997) Remote estimation of chlorophyll content in higher plant leaves. Int J Remote Sens 18: 2691-2697
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Großkinsky DK, Svensgaard J, Christensen S, Roitsch T (2015) Plant phenomics and the need for physiological phenotyping
across scales to narrow the genotype-to-phenotype knowledge gap. J Exp Bot 66: 5429-5440
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Hansen PM, Schjoerring JK (2003) Reflectance measurement of canopy biomass and nitrogen status in wheat crops using
normalized difference vegetation indices and partial least squares regression. Remote Sens Environ 86: 542-553
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Herrmann I, Karnieli A, Bonfil DJ, Cohen Y, Alchanatis V (2010) SWIR-based spectral indices for assessing nitrogen content in
potato fields. Int J Remote Sens 31: 5127-5143
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Hewitt EJ, Smith TA (1975) Plant mineral nutrition. English Universities Press Ltd., London.
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Irisarri JGN, Oesterheld M, Verón SR, Paruelo JM (2009) Grass species differentiation through canopy hyperspectral reflectance.
Int J Remote Sens 22: 5959-5975
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Jacquemoud S, Baret F (1990) PROSPECT: a model of leaf optimal properties spectra. Remote Sens Environ 34: 75-91
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Jones MGK, Outlaw WH, Lowry OH (1977) Enzymic assay of 10-7 to 10-14 moles of sucrose in plant tissues. Plant Physiol 60: 379383
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Kraft M, Weigel H-J, Mejer G-J, Brandes F (1996) Reflectance measurements of leaves for detecting visible and non-visible
damage to crops. J Plant Physiol 148: 148-154
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Kim KM, Kwon YS, Lee JJ, Eun MY, Sohn JK (2004) QTL mapping and molecular marker analysis for the resistance of rice to
ozone. Molecules Cells 17: 151-155
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
le Maire G, Francois C, Soudani K, Berveiller D, Pontailler JY, Breda N, Genet H, Davi H, Dufrene E (2008) Calibration and
validation of hyperspectral indices for the estimation of broadleaved forest leaf chlorophyll content, leaf mass per area, leaf area
index and leaf canopy biomass. Remote Sens Environ 112: 3846-3864
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
le Maire G, Francois C, Dufrene E (2004) Towards universal broad leaf chlorophyll indices using PROSPECT simulated database
and hyperspectral reflectance measurements. Remote Sens Environ 89: 1-28
Pubmed: Author and Title
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Leitao L, Bethenod O, Biolley JP (2007) The impact of ozone on juvenile maize (Zea mays L.) plant photosynthesis: effects on
vegetative biomass, pigmentation, and carboxylases (PEPc and Rubisco). Plant Biol 9: 478-488
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Lichtenthaler HK, Wellburn AR (1983) Determination of total carotenoids and chlorophylls a and b of leaf extracts in different
solvents. Biochem Soc Trans 11: 591-592
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Liu L, Cheng Z (2011) Mapping C3 and C4 plant functional types using separated solar-induced chlorophyll fluorescence from
hyperspectral data. Int J Remote Sens 32: 9171-9183
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Maccioni A, Agati G, Mazzinghi P (2001) New vegetation indices for remote measurement of chlorophylls based on leaf directional
reflectance spectra. J Photochem Photobiol B 61: 52-61
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Markelz RJC, Strellner RS, Leakey ADB (2011) Impairment of C4 photosynthesis by drought is exacerbated by limiting nitrogen and
ameliorated by elevated [CO2] in maize. J Exp Bot 62: 3235-3246
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Martin ME, Aber JD (1997) High spectral resolution remote sensing of forest canopy lignin, nitrogen and ecosystem processes.
Ecol Appl 7: 431-443
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
McGrath JM, Betzelberger AM, Wang S, Shook E, Zhu XG, Long SP, Ainsworth EA (2015) An analysis of ozone damage to historical
maize and soybean yields in the United States. Proc Natl Acad Sci USA 112: 14390-14395
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
McMullen MD, Kresovich S, Villeda HS, Bradbury P, Li HH, et al. (2009) Genetic properties of the maize nested association
mapping population. Science 325: 737-740
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
McMurtey JE III, Chapelle EW, Kim MS, Meisinger JJ, Corp LA (1994) Distinguish nitrogen fertilization levels in field corns (Zea
mays L.) with actively induced fluorescence and passive reflectance measurements. Remote Sens Environ 47: 36-44
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Meroni M, Panigada C, Rossini M, Picchi V, Cogliati S, Colombo R (2009) Using optical remote sensing techniques to track the
development of ozone-induced stress. Environ Pollut 157: 1413-1420
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Mevik BH, Wehrens R (2007) The pls package: principal component and partial least squares regression in R. J Stat Software 18:
1-24
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Miller JD, Arteca RN, Pell EJ (1999) Senescence-associated gene expression during ozone-induced leaf senescence in
Arabidopsis. Plant Physiol 120: 1015-1024
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Morgan PB, Bernacchi CJ, Ort DR, Long SP (2004) An in vivo analysis of the effect of season-long open-air elevation of ozone to
anticipated 20150 levels on photosynthesis in soybean. Plant Physiol 135: 2348-2357
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
Penuelas J, Baret F, Filella I (1995) Semi-empirical indices to assess carotenoids/chlorophyll a ratio from leaf spectral reflectance.
Photosynthetica 31: 221-230
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Penuelas J, Filella I (1998) Visible and near-infrared reflectance techniques for diagnosing plant physiological status. Trends Plant
Sci 3: 151-156
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Poorter H, Niklas KJ, Reich PB, Oleksyn J, Poot P, Mommer L (2012) Biomass allocation to leaves, stems and roots: meta-analyses
of interspecific variation and environmental control. New Phytol 193: 30-50
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Reich PB, Amundson RG (1985) Ambient levels of ozone reduce net photosynthesis in tree and crop species. Science 230: 566-570
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Reich PB, Walters MB, Ellsworth DS (1997) From tropics to tundra: global convergence in plant functioning. Proc Natl Acad Sci
USA 94: 13730-13734
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Richardson AD, Reeves JB (2005) Quantitative reflectance spectroscopy as an alternative to traditional wet lab analysis of foliar
chemistry: near-infrared and mid-infrared calibrations compared. Can J For Res 35: 1122-1130
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Rudorff BFT, Mulchi CL, Daughtry CST, Lee EH (1996) Growth, radiation use efficiency, and canopy reflectance of wheat and corn
grown under elevated ozone and carbon dioxide atmospheres. Remote Sens Environ 55: 163-173
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Runeckles VC, Resh HM (1975) The assessment of chronic ozone injury to leaves by reflectance spectrophotometry. Atmos
Environ 9: 447-452
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Serbin SP, Dillaway DN, Kruger EL, Townsend PA (2012) Leaf optical properties reflect variation in photosynthetic metabolism and
its sensitivity to temperature. J Exp Bot 63: 489-502
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Serbin SP, Singh A, Desai AR, Dubois SG, Jablonski AD, Kingdon CC, Kruger EL, Townsend PA (2015) Remotely estimating
photosynthetic capacity, and its response to temperature, in vegetation canopies using imaging spectroscopy. Remote Sens
Environ 167: 78-87
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Serbin, SP, Singh A, McNeil BE, Kingdon CC, Townsend PA (2014) Spectroscopic determination of leaf morphological and
biochemical traits for northern temperate and boreal tree species. Ecol Appl 24: 1651-1669.
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Serrano L, Penuelas J, Ustin SL (2002) Remote sensing of nitrogen and lignin in Mediterranean vegetation from AVIRIS data:
decomposing biochemical from structural signals. Remote Sens Environ 81: 355-364
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Siegmann B, Jarmer T (2015) Comparison of different regression models and validation techniques for the assessment of wheat
leaf area index from hyperspectral data. Int J Remote Sens 36: 4519-4534
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Sims DA, Gamon JA (2002) Relationships between leaf pigment content and spectral reflectance across a wide range of species,
leaf structures and developmental stages. Remote Sens Environ 81: 337-354
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Street NR, Tallis MJ, Tucker J, Brosché M, Kangasjärvi J, Broadmeadow M, Taylor G (2011) The physiological, transcriptional and
genetic responses of an ozone-sensitive and an ozone tolerant poplar and selected extremes of their F2 progeny. Environ Pollut
159: 45-54
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Tilman D, Balzer C, Hill J, Befort BL (2011) Global food demand and the sustainable intensification of agriculture. Proc Natl Acad
Sci USA108: 20260-20264
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Ueda Y, Siddique S, Frei M (2015) A novel gene, OZONE-RESPONSIVE APOPLASTIC PROTEIN1, enhances cell death in ozone
stress in rice. Plant Physiol 169: 873-889
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Vogelman JE, Rock BN, Moss DM (1993) Red edge spectral measurements from sugar maple leaves. Int J Remote Sens 14: 15631575
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
von Caemmerer S (2000) Biochemical models of leaf photosynthesis. CSIRO Publishing, Collingwood, Victoria.
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Weber VS, Araus JL, Cairns JE, Sanchez C, Melchinger AE, Orsini E (2012) Prediction of grain yield using reflectance spectra of
canopy and leaves in maize plants grown under different water regimes. Field Crops Res 128: 82-90.
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Wold S, Sjostrom M, Eriksson L (2001) PLS-regression: a basic tool of chemometrics. Chemometr Intell Lab 58: 109-130
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Wold S (1994) PLS for multivariate linear modeling. In van de Waterbeernd H, editor. Chemometric methods in molecular design,
methods and principles in medicinal chemistry. Verlag-Chemie, Weinheim, Germany, pp 195-218
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Wright IJ, Reich PB, Westoby M, Ackerly DD, Baruch Z, Bongers F, Cavender-Bares J, Chapin R, Cornelissen JHC, Diemer M, et al.
(2004) The worldwide leaf economics spectrum. Nature 428: 821-827
Pubmed: Author and Title
CrossRef: Author and Title
Google Scholar: Author Only Title Only Author and Title
Downloaded from on June 18, 2017 - Published by www.plantphysiol.org
Copyright © 2016 American Society of Plant Biologists. All rights reserved.