| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Centre UdL-IRTA, Alcalde Rovira Roure 177, 25198 Lleida, Spain
b Servei de Camps Experimentals, Barcelona, Spain
c Unitat de Fisiologia Vegetal, Facultat de Biologia, Univ. de Barcelona, Diagonal 645, 08028 Barcelona, Spain
conxita.royo{at}irta.es
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
Abbreviations: CDM, crop dry mass • GAI, green area index • GAP, green area per plant • GDD, growing degree-days • LAI, leaf area index • NDVI, normalized difference vegetation index • NIR, near-infrared radiation • PAR, photosynthetically active radiation • PRI, photochemical reflectance index • RUE, photosynthetic radiation-use efficiency • SR, simple ratio. **, Significant at the 0.01 probability level
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
The use of morphological and physiological traits as indirect selection criteria for grain yield is an alternative breeding approach. This has come to be known as analytical breeding (Richards, 1982), and implies a better understanding of the factors controlling development, growth, and grain yield (Shorter et al., 1991). In recent years, many selection criteria based on morphological, physiological, and biochemical traits have been suggested (see, for example, in relation to durum wheat: Dib et al., 1994; Loss and Siddique, 1994; Nachit et al., 1992), but rarely adopted in any breeding program. The limited success to date of the analytical approach may be due to a lack of understanding of the physiological factors most directly involved in determining yield, in addition to the absence of proper methods for evaluating them in a rapid, routine manner (Blum, 1988, p. 223; Loss and Siddique, 1994; Richards, 1996). Among the most promising screening methods are those that allow a quick screening of physiological traits that can integrate the performance of the crop either over time (i.e., during plant cycle) or at the organizational level (i.e., whole plant, canopy) (Araus, 1996; Richards, 1996).
While productivity is determined by reference to many processes, the factors determining yield can be divided, in a simple approach, into a small number of integrative physiological traits. For example, the potential yield of a crop over a given period of time and under particular growing conditions is determined by three major processes or integrative traits (Hay and Walker, 1989, p. 250): first, the interception of incident solar irradiance by the canopy; second, the conversion of the intercepted radiant energy to potential chemical energy; and third, the harvest index. Whereas harvest index has to date been the only trait successfully used to improve yield primarily under optimum conditions (Slafer et al., 1993, and references therein), the other two processes, which are responsible for the overall crop biomass, are particularly affected by drought and other related stresses (Sinclair, 1988).
The first of these processes (interception of incident solar radiation) depends on the photosynthetic area of the canopy, while the second (conversion to potential chemical energy) relies on the overall photosynthetic efficiency of the crop. The measurement of spectra reflected by crop canopies at different wavelengths through the photosynthetically active radiation (PAR) and near-infrared radiation (NIR) regions of the electromagnetic spectrum can estimate simultaneously, and in a rapid nondestructive manner, photosynthetic traits such as canopy green area and radiation-use efficiency (RUE), important attributes in determining yield (Araus, 1996; Field et al., 1994; Peñuelas, 1998). Such estimations are made using spectral reflectance indices, which are formulations based on simple operations between the reflectances at given wavelengths (e.g., ratios and differences).
The most widespread application of these indices is in the assessment of those characteristics related to the total photosynthetic area of the canopy. The most widely used spectral vegetation indices are the SR and the NDVI (see Materials and Methods for their calculation). These indices have been correlated positively (either on a linear or a logarithmic basis) with CDM, LAI, GAI, and the fraction of PAR radiation absorbed by the canopy in small-grain cereals such as bread wheat and barley (Hordeum vulgare L.) (Bellairs et al., 1996; Fernández et al., 1994; Field et al., 1994; Peñuelas et al., 1997). However, these studies have been carried out in a small number of genotypes, which restricts the use of such techniques for breeding purposes. Besides, no results have been published on durum wheat until now (Carlson and Ripley, 1997; Peñuelas, 1998). Nevertheless, the ordinary methods of measuring biomass in cereal plots involve a destructive and tedious sampling, which is not suitable for routine assessment in plant breeding. In this context, the spectral vegetation indices have the potential to provide a nondestructive and fast estimate of plant biomass production. Moreover, the photosynthetic radiation-use efficiency (RUE, the second component in the above identity) of the PAR absorbed by the canopy has been correlated with the PRI (Filella et al., 1996; Gamon et al., 1992, 1997; Peñuelas et al., 1994, 1995). This highlights the potential interest of spectroradiometrical techniques for agronomical and plant breeding purposes.
Although spectral reflectance indices have been used to assess yield in cereals such as bread wheat, barley, rice (Oryza satira L.), maize (Zea mays L.), and sorghum [Sorghum bicolor (L.) Moench], information on the performance of these indices for durum wheat is needed. Such assessment is particularly relevant under Mediterranean conditions, where most of the durum wheat is grown. In addition, the lack until recently of field portable spectroradiometers has restricted the application of this technique to breeding programs, where the routine assessment of large amounts of germplasm is necessary. In this context, our objective was to study the performance of the NDVI, SR, and PRI, assessing changes (during the second half of the crop cycle) in biomass and green area, as well as in the final yield of durum wheat. Since drought stress is the main constraint on durum wheat production in Mediterranean conditions, the evaluation of these spectral reflectance indices was carried out, in this study, in environments with contrasting water availability. As Wiegand et al. (1991) point out, if most uncertainty in yield prediction by spectral reflectance indices is site-dependent, then it is necessary to relate yield vs. these indices across good and poor environments within the production area. In addition, Daughtry et al. (1983) suggested that when the correlation between grain yield and spectral indices is obtained on a single observation date, it must be used with caution. Therefore, the additive effect of the vegetative indices (SR and NDVI) during the crop cycle, which account for differences in yield, was further evaluated. Indeed, it has been reported that yield can be calculated from successive measurements of spectral vegetation indices during the growing season (Daughtry et al., 1992; Rudorff and Batista, 1990; Wiegand and Richardson, 1990; Wiegand et al., 1991).
|
Materials and methods |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
Twenty-five durum wheat genotypes, including four Spanish commercial cultivars (Altar-Aos, Jabato, Mexa, and Vitrón), and 21 cultivars and advanced lines from the International Maize and Wheat Improvement Center (CIMMYT)–International Center for Agricultural Research in the Dry Areas (ICARDA) durum wheat breeding program (Awalbit, Bicrecham-1, Chacan, Chahra-1, Haurani, Korifla, Krs/Haucan, Lagost-3, Lahn/Haucan, Massara-1, Moulchahba-1, Mousabil-2, Omlahn-3, Omrabi-3, Omruf-3, Quadalete//Erp/Mal, Sebah, Stojocri-3, Waha, Zeina-1, and Zeina-2), representing a wide range of genetic variability, were sown in an 
-lattice design, with four replicates. Both trials were sown on 3 Dec. 1996 in 12-m2 plots (six rows, 20 cm apart). The experimental plots were divided into two 5-m-long subplots. One of these was used for successive destructive samplings of biomass; the other half remained intact for radiometric and grain yield determinations. Sowing density was 550 viable seeds m-2 .
Before sowing, the irrigated trial was fertilized with 50 kg N ha-1, 120 kg P ha-1, and 165 kg K ha-1 and the rainfed trial with 32 kg N ha-1, 60 kg P ha-1, and 60 kg K ha-1. The trials were top-dressed at the onset of jointing with 100 kg N ha-1 at the irrigated site and with 84 kg N ha-1 at the rainfed site. Climatic data during the growing season are summarized in Table 1 . The water available for the crop from seedling to the onset of jointing (mid-March, when irrigation began) was similar under both irrigated and rainfed environments. Under irrigated conditions, the crop was flooded–irrigated three times (50 mm of water on each occasion) at monthly intervals during the spring. The total amount of water available for the crop was 462 and 257 mm for irrigated and rainfed trials, respectively.
|
Growth Indices and Grain Yield
Zadoks stage was determined weekly for each plot. For biomass determinations, at each sampling the plants contained in 0.5 m row length (40–50 plants) were pulled up at random from a central row in each plot, and five representative plants per plot were taken at random and separated into leaves, culms and spikes. The leaf area (one side) was measured using a leaf-area meter (DIAS II, Delta-T Devices, Cambridge, UK). Yellow and dry leaves were excluded from the measurement. The samples were oven-dried at 80°C for 48 h and weighed. Crop dry mass (g m-2), green LAI, and total GAI were calculated for each plot as follows:
, where N = number of plants m-2 of soil and M = mass (g of the aerial part) plant-1; LAI = N x LAP, where LAP = leaf area plant-1, and GAI = N x GAP, where GAP = green (leaf plus culm and projection of the spike) area plant-1. Thermal time was calculated in growing degree-days (GDD) by summing the daily values of mean temperature, with a base temperature of 0°C (Gallagher, 1979).
Plots were mechanically harvested on 8 July and 25 June 1997 for irrigated and rainfed trials, respectively. Grain yield (kg ha-1) was determinated on a plot basis, and is reported at a 100 g kg-1 moisture level.
Spectral Reflectance Measurements
Canopy reflectance was detected with a narrow-bandwidth visible–near-infrared portable field spectroradiometer fitted with an 18° field-of-view optic (Model FieldSpec UV/VNIR, Analytical Spectral Devices, Boulder, CO). The instrument detects 512 continuous bands (with a sampling interval of 1.4 nm) from 350- to 1050-nm wavelengths, thereby covering the visible and near-infrared portion of the spectrum. Individual scans were remotely triggered from a portable computer and saved to disk for subsequent analysis. The measurements were made at midday under cloudless conditions. All spectral measurements were taken with the sensor at a zenith angle of 60°, with the field of view optic mounted on a tripod 1.5 m above the soil and with the radiometer in a nadir orientation. Three spectral reflectance measurements (1–2 s each) were taken at each plot, each being the average of five scans. The reflectance spectrum was calculated in real time as the ratio between the reflected and the incident spectra on the canopy, where the incident spectrum was periodically obtained from the light reflected by a white reference panel (Spectralon, Labsphere, North Sutton, NH). The Spectralon is very close to a lambertian surface. Reflectance standard measurements were made on each of the five plots. Further, radiometric indices were calculated from spectral reflectance measurements. Vegetation indices were calculated from the comparison between visible and near-infrared reflectance. Thus, the NDVI was calculated as (R900 - R680)/(R900 + R680) and the SR as R900/R680, following Peñuelas et al. (1997). The PRI is given as (R531 - R570)/(R531 + R570) (Gamon et al., 1992; Peñuelas et al., 1995). In all cases, Rn is the reflectance at the wavelength (in nm) indicated by the subscript. The reflectance values used in the calculation of SR, NDVI, and PRI include reflection by the canopy as well as reflection by the soil background. The reflection spectrum of the soil background was homogeneous in each site and similar in the two sites.
Statistical Analysis
-Lattice analysis was used to analyze biomass, reflectance indices, and grain yield data; the least square means were obtained by the MIXED procedure of the SAS/STAT statistical package (SAS Inst., 1987). Pearson correlation coefficients were used to study the relationship between radiometric indices and biological variables. The percentage of grain yield variation explained by the progressive addition of each spectral reflectance index measured at different growth stages was assessed by means of the coefficient of determination of a multilineal fitting.
|
Results |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
|
Effect of Environment and Ontogeny on Crop Growth and Spectral Reflectance Indices
All the crop growth indices (CDM, LAI, and GAI) and spectral reflectance indices (NDVI, SR, and PRI) studied were higher under irrigated than under rainfed conditions. However, significant G x E interactions appeared for some of them. Except for CDM, the differences between environments for all these parameters were already significant at booting and remained so until maturity (Table 3)
.
|
Simple ratio in both environments showed a fairly similar pattern to that of LAI during grain filling, decreasing progressively after anthesis (Table 3). The pattern of changes in SR from booting to anthesis resembled closely those of GAI; thus, while SR increased under irrigation, it decreased under rainfed conditions. Normalized difference vegetation index also diminished during grain filling, although more slowly, and showed the same pattern of changes as SR, although not to such an extreme extent. Photochemical reflectance index decreased in both environments from anthesis to maturity.
Relationship between Crop Growth and Spectral Reflectance Indices
Significant (P < 0.05) positive correlations between CDM and the vegetation indices NDVI and SR were observed only at milk-grain stage in the irrigated environment and at physiological maturity in the rainfed environment (Table 4)
. Crop dry mass was positively correlated with PRI only at physiological maturity under irrigation. Spectral reflectance indices correlated better with the photosynthetic (leaf and total) area of the canopy than with CDM, particularly in the rainfed environment. In this environment, LAI, and to a slightly lesser extent GAI, showed significant positive correlations with SR and NDVI (and to a lesser degree with PRI) measured at any of the growth stages assayed. In the irrigated environment, the relationships were lower and only significant during the second half of the grain filling period.
|
|
|
|
|
|
Discussion |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
680 nm), and a consequent blue shift on the red edge (Gamon et al., 1992; Peñuelas et al., 1997). Such changes in the spectral signature lead to a decrease in either NDVI and SR. Ontogenetic changes in the spectral vegetation indices also reflect the pattern of changes during the crop cycle in green area (see Table 3). Differences in the PRI were produced by a differential reflectance in the zeaxanthin carotenoid active band (531 nm) (Gamon et al., 1992; Peñuelas et al., 1995), which suggests that the RUE of plants was higher in the irrigated than in the rainfed environment (Filella et al., 1996). Similarly, the reduction that occurred in the two environments in PRI from anthesis to maturity was probably associated with a progressive senescence of photosynthetic organs during grain filling. The weaker relationships recorded between the spectral reflectance indices and GAI, compared with those for LAI, might have been the result of the difficulties encountered in measuring properly the green area of organs other than leaf laminae (e.g., the spike and the stem). Similarly, problems associated with the particular architecture of these nonlaminar green organs when spectral reflectance indices are measured should not be forgotten. In wheat, the relationship between spectral vegetation indices and canopy parameters has been reported as being disturbed by the presence of the spike (Shibayama et al., 1986).
The exponential pattern of the relationship between NDVI and LAI (Fig. 1b) agrees with that reported previously in bread wheat (Curran, 1983), where an LAI value of 3 for the beginning of the asymptotic region was found. Indeed, the NDVI is a highly sensitive index for crops with LAI between 0 and 2. For crops with LAI > 3, the addition of more canopy layers has little effect on the relative interceptance or reflectance of red and near-infrared radiation, and thus little effect on difference to NDVI (Sellers, 1987). This in itself suggests a potential limitation of the use of NDVI as a crop area indicator, since beyond a certain value of LAI the changes in NDVI with LAI become insignificant (Carlson and Ripley, 1997). The NDVI is ideally suited for detecting subtle differences in cover in sparse canopies, however, and as such constitutes a sensitive growth index either in early crop stages or under stress conditions (Bellairs et al., 1996; Gallo et al., 1985; Peñuelas et al., 1997). Similarly, our results (see the initial part of the relationship in Fig. 1b) with durum wheat indicate that, even under adequate growing conditions, NDVI may also be useful in the later crop stages (i.e., during grain filling), when LAI decreases to values around 2 (Table 3). This observation agrees with that previously reported in bread wheat under Mediterranean conditions by Fernández et al. (1994). Therefore, the pattern of the relationship between NDVI and LAI supports the better performance of NDVI in the rainfed environment than under irrigation when assessing differences in green area and grain yield. In addition, other factors such as the kind of vegetation, developmental stage, and leaf water content may be involved (Paltridge and Barber, 1988).
Among the vegetation indices, SR correlated better with crop growth (measured as total biomass or photosynthetic area) and grain yield than NDVI, specially under rainfed conditions. The linear nature of the relationship between SR and LAI (Fig. 1a) compared with the exponential relationship between NDVI and LAI (Fig. 1b) supports this fact. This linear relationship between SR and LAI has been widely reported (see references in Araus, 1996; Field et al., 1994; Peñuelas, 1998). Nevertheless, under irrigation, the SR values for LAI beyond 2.5 were highly dispersed (Fig. 1a), which is consistent with the absence in this environment of significant correlations between SR and LAI at heading and anthesis (Table 4). At increasing LAI, due to the algebraic definition of each vegetation index, NDVI would tend asymptotically to 1, while SR would tend to infinity. However, due to saturation of the reflectance, vegetation index becomes insensitive at LAI 3. At higher LAIs, therefore, SR shows large scattering due to the noise of the measurement, while NDVI fluctuates slightly below 1.
Photochemical reflectance index showed an exponential relationship pattern with LAI (Fig. 1c). This may be due to the normalized nature of this index, which has a mathematical formula similar to that of NDVI. The weaker correlations of PRI with grain yield compared with those of SR and NDVI may be explained by the fact that any reduction in the RUE (due to drought and associated stresses or just to plant ontogeny) is generally less significant than that in total biomass or intercepted PAR (Sinclair, 1988). Alternatively, the relationship between PRI and green area, as well as the absence in PRI of any effect independent of that of SR or NDVI to explain differences in yield, suggests that PRI assesses similar characteristics of the crop to those that are assessed by vegetation indices. Thus, for example, during grain filling, the effect of drought and plant ontogeny on senescence will simultaneously affect leaf area duration and the RUE.
In conclusion, our results, based on a wide range of genetic variability, suggest that for durum wheat under Mediterranean conditions, the usefulness of SR and NDVI for predicting green area and grain yield is limited to environments or crop stages in which the LAI values are <3 (Fig. 3). Thus, for example, in the semiarid conditions typical of rainfed Mediterranean environments, a major breeding trait such as the LAI attained at anthesis (Loss and Siddique, 1994) could be assessed by means of these vegetation indices. Indeed, in the rainfed environment, the best correlation between these spectral reflectance indices and yield was attained at anthesis. In the more favorable irrigation environment (with an LAI > 3 for most of the crop cycle), the usefulness of these indices is less evident and restricted to the later stages of the crop. Indeed, our results for the rainfed environment concerning the percentage variability in yield across genotypes explained by the integration of the spectral vegetation indices during the crop cycle agree with previous results in bread wheat (Fig. 4). Thus, Wiegand and Richardson (1990) reported an r2 of 0.50 for the prediction of grain yield from spectral vegetation indices measured at four dates during vegetative growth. Similarly, Rudorff and Batista (1990) reported an r2 of 0.66 between yield and integrated spectral vegetation indices from booting to completely senesced plants. The present study extends to durum wheat the potential validity of spectral vegetation indices in assessing genotypic differences in yield and biomass under Mediterranean conditions, especially for environments with low and medium yield potential.SAS Institute 1987
|
ACKNOWLEDGMENTS |
|---|
Received for publication January 4, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMaterials and methodsResultsDiscussionREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. S. Carley, D. L. Jordan, L. C. Dharmasri, T. B. Sutton, R. L. Brandenburg, and M. G. Burton Peanut Response to Planting Date and Potential of Canopy Reflectance as an Indicator of Pod Maturation Agron. J., February 26, 2008; 100(2): 376 - 380. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Prasad, B. F. Carver, M. L. Stone, M. A. Babar, W. R. Raun, and A. R. Klatt Genetic Analysis of Indirect Selection for Winter Wheat Grain Yield Using Spectral Reflectance Indices Crop Sci., July 30, 2007; 47(4): 1416 - 1425. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Prasad, B. F. Carver, M. L. Stone, M. A. Babar, W. R. Raun, and A. R. Klatt Potential Use of Spectral Reflectance Indices as a Selection Tool for Grain Yield in Winter Wheat under Great Plains Conditions Crop Sci., July 30, 2007; 47(4): 1426 - 1440. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. G. Sullivan and C. C. Holbrook Using Ground-Based Reflectance Measurements as Selection Criteria for Drought- and Aflatoxin-Resistant Peanut Genotypes Crop Sci., May 31, 2007; 47(3): 1040 - 1050. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. Board, V. Maka, R. Price, D. Knight, and M. E. Baur Development of Vegetation Indices for Identifying Insect Infestations in Soybean Agron. J., April 4, 2007; 99(3): 650 - 656. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Babar, M. P. Reynolds, M. van Ginkel, A. R. Klatt, W. R. Raun, and M. L. Stone Spectral Reflectance to Estimate Genetic Variation for In-Season Biomass, Leaf Chlorophyll, and Canopy Temperature in Wheat Crop Sci., March 27, 2006; 46(3): 1046 - 1057. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. A. Babar, M. P. Reynolds, M. van Ginkel, A. R. Klatt, W. R. Raun, and M. L. Stone Spectral Reflectance Indices as a Potential Indirect Selection Criteria for Wheat Yield under Irrigation Crop Sci., February 1, 2006; 46(2): 578 - 588. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. Board and H. Modali Dry Matter Accumulation Predictors for Optimal Yield in Soybean Crop Sci., August 1, 2005; 45(5): 1790 - 1799. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K.-W. Chang, Y. Shen, and J.-C. Lo Predicting Rice Yield Using Canopy Reflectance Measured at Booting Stage Agron. J., May 13, 2005; 97(3): 872 - 878. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. L. Ma, K. D. Subedi, and C. Costa Comparison of Crop-Based Indicators with Soil Nitrate Test for Corn Nitrogen Requirement Agron. J., March 1, 2005; 97(2): 462 - 471. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Aparicio, D. Villegas, J. L. Araus, J. Casadesus, and C. Royo Relationship between Growth Traits and Spectral Vegetation Indices in Durum Wheat Crop Sci., September 1, 2002; 42(5): 1547 - 1555. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. L. ARAUS, G. A. SLAFER, M. P. REYNOLDS, and C. ROYO Plant Breeding and Drought in C3 Cereals: What Should We Breed For? Ann. Bot., June 15, 2002; 89(7): 925 - 940. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. F. Shanahan, J. S. Schepers, D. D. Francis, G. E. Varvel, W. W. Wilhelm, J. M. Tringe, M. R. Schlemmer, and D. J. Major Use of Remote-Sensing Imagery to Estimate Corn Grain Yield Agron. J., May 1, 2001; 93(3): 583 - 589. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Crop Science | Vadose Zone Journal | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
