Open Collections

UBC Undergraduate Research

Evaluating the use of near-infrared reflectance spectroscopy as a proxy measure of carbon and nitrogen… Leung, Andy 2014-04-07

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
52966-Leung_Andy_FRST_498.pdf [ 444.38kB ]
Metadata
JSON: 52966-1.0075640.json
JSON-LD: 52966-1.0075640-ld.json
RDF/XML (Pretty): 52966-1.0075640-rdf.xml
RDF/JSON: 52966-1.0075640-rdf.json
Turtle: 52966-1.0075640-turtle.txt
N-Triples: 52966-1.0075640-rdf-ntriples.txt
Original Record: 52966-1.0075640-source.json
Full Text
52966-1.0075640-fulltext.txt
Citation
52966-1.0075640.ris

Full Text

     Evaluating the use of near-infrared reflectance spectroscopy as a proxy measure of carbon and nitrogen isotopes in the leaves of Black Cottonwood (Populus trichocarpa)     by   Andy Leung   ii  Abstract  Stable isotopic measurements are one of the more powerful tools used to help advance our understating of plants and their environment. Yet this tool is underutilized because of the large amount of resources and time it takes to extract this information. In this study, I evaluated if near-infrared reflectance spectroscopy can be used as a faster and more economical way to estimate the ratios of carbon and nitrogen isotopes in tree leaves. The δ13C and δ15N values were determined with samples of Black Cottonwood leaves (Populus trichocarpa) taken from 3 different clones and grown in two different CO2 concentration conditions. The samples were scanned with a near-infrared reflectance spectrometer (NIRS) to create calibration models. These models are created using partial least-squares regressions and tested by cross validation procedures. The resulting calibration models were unable to accurately predict the amount of δ13C and δ15N in the leaves as none of the models could produce a high correlation coefficient. The reflectance spectra produced by the NIRS was able to differentiate the two different CO2 concentration treatments, and was also able to classify clones from different origins based on their reaction to the CO2 concentration treatments.    Key words: δ13C, δ15N, isotopic composition, leaf tissue, near-infrared reflectance spectroscopy, partial least-squares regression. iii  Abstract............................................................................................................................................ ii Table of Contents Table of Contents ........................................................................................................................... iii Index of Tables ............................................................................................................................... iii Index of Figures  ............................................................................................................................. iii Introduction  ................................................................................................................................... 1 Methods  ......................................................................................................................................... 3 Results  ............................................................................................................................................ 5 Discussion  .................................................................................................................................... 13 References  ................................................................................................................................... 16 Index of Tables Table 1: NIR calibration and cross validation statistics of PLSR models .................................. 5 Table 2: Statistical summary of the sample set ................................................................. 6 Index of Figures Figure 1: Relationship between NIRS predicted values and measurements by reference methods. For prediction, the partial least-square regression (PLSR) calibration model with the best cross validation statistics is presented.  (a) Predicting δ13C [‰] using all the samples.  (b) Predicting δ15N [‰] using all the samples.  (c) Predicting δ13C [‰] using ambient CO2 treated samples only.  (d) Predicting δ15N [‰] using ambient CO2 treated samples only.  (e) Predicting δ13C [‰] using enriched CO2 treated samples only.  (f) Predicting δ15N [‰] using enriched CO2 treated samples only. ......................................... 7 Figure 2: Ambient / Enriched CO2 treatment changes in reflectance spectra graphs  (a) Average ambient CO2 treated samples / average enriched CO2 treated samples.  (b) Ambient / enriched CO2 treated samples from the Bell-Irving River clone replicates (I1).  (c) Ambient / enriched CO2 treated samples from the Bell-Irving River clone replicates (I1L).  (d) Ambient / enriched CO2 treated samples from the Jasper River clone replicates (J1).  (e) Ambient / enriched CO2 treated samples from the Jasper River clone replicates (J2).  (f) Ambient / enriched CO2 treated samples from the Jasper River clone replicates (J4).  (g) Ambient / enriched CO2 treated samples from Quesnel River clone replicates.  ............ 101  Introduction  Stable isotope methods have recently emerged as one of the more powerful tools for advancing the understanding of relationships between plants and their environment (Dawson et al., 2002).They are used extensively to examine physiological, ecological, and biogeochemical processes and provide information at a variety of temporal and spatial scales (Bowling et al., 2008). Isotopes with higher mass are usually discriminated against in a reaction because they are less reactive than their lighter counterpart due to the strength of the bonds, which will cause isotopic fractionation between substrate and the product (Farquhar et al., 1989). For example, understanding the ratio of carbon 13 in relation to total carbon in plants cells has been used to classify higher plants into the 3 photosynthetic modes: C3, C4 & CAM (Farquhar et al., 1989). The isotopic ratio of carbon 13 can also be used to assess water-use efficiency of conifers trees and other plants with different environmentally induced and genetic differences (e.g., Sun et al., 1996).  Traditionally, determining the variation in isotopic composition of a sample requires expensive and not very portable instrumentation like an isotope ratio mass spectrometer. Also samples are destroyed after the analyses are done and analysis can be time-consuming.  Because of these limitations, many workers are not able to use isotopic composition as one of their measurements for rapid screening of a large amount of samples to detect interesting patterns (Kleinebecker et al., 2009). As technology develops, near-infrared reflectance spectroscopy (NIRS) has been introduced as a fast low-cost alternative for analysis of the chemical composition of organic materials, especially in the food industry (Clark et al., 1987). A near-infrared spectrometer is very portable compared to the mass spectrometer and costs a fraction of the price. But the biggest advantages of NIRS over mass spectrometry are the speed of analysis, there is little or no sample preparation and that the sample is not burnt or used up after the analyses. Near-infrared radiation is use to induce vibration between C-H, N-H, and O-H bonds which will reflect a different amount of radiation back at different frequency resulting a spectrum (Shenk et al., 2008). To predict the composition of 2  unknown samples, a model is then made from these spectrums of samples with known chemical composition (Workman, 2008). Work has been done by Kleinebecker et al. (2009) to show that there is a strong correlation between predicted versus actual values of carbon 13 and nitrogen 15 isotopes using the calibration models they constructed with leaves from seven bog species from southern Patagonia. The goal of my project is to see if there are such correlations between predicted versus actual values of carbon 13 and nitrogen in Black Cottonwood (Populus trichocarpa) leaves. Towards this goal, I had also attempted to isolate the specific part of the reflectance spectrum responsible for carbon 13 and nitrogen 15 isotopes.   3  Methods The Populus trichocarpa leaf samples used in this analysis was collected from a previous experiment done by Buschhaus (2007) on the tissue δ15N of Populus trichocarpa grown in steady-state NH4+ nutrition. Three different clones from each latitudinally dispersed populations of Populus trichocarpa from Jasper River, OR (44oN), Quesnel River, BC (52oN) and Bell-Irving River, BC (56oN) were chosen for the experiment. Uniform, 5 cm cuttings were rooted and grown in hydroponic medium containing 400 µM of (NH4)SO4. Half of the samples were then grown in a growth chamber with ambient CO2 concentration of 400 µL L-1 while the other samples were grown in a different chamber with all the same nutrients except with enriched CO2 concentration of 800 µL L-1 inside. The tank CO2 used for these experiments was depleted in 13C relative to normal air by ~24‰, ensuring a large difference in isotopic composition of the plants. After 6 weeks of growth, leaves of each sample were collected and freeze-dried before being ground into fine powder using a ball-mill (Fritsch Laborgerätebau, Terochem Scientific).  The δ13C[‰] and δ15N[‰] values of each leaf sample were determined on a Europa ANCA-SL preparation module and a Europa Hydra 20/20 isotope ratio mass spectrometer (University of California Stable Isotope Facility, Davis, CA). There were four characters on the label of each samples. The first two characters identified the clone's origin. The last two characters were the unique identifier for each replicate.    The leftover fine powder samples from the previous analysis was then stored in the laboratory at room temperature and humidity. Samples were then dried by keeping them in air-tight containers partly filled with desiccant for three days before the NIRS analysis (this was found to be essential to obtain good reproducibility). Each sample was then manually packed into the sample capsule with a quartz glass cover and scanned with a QualitySpec Pro Vis/NIR Spectrometer (Analytical Spectral Devices, Inc., Boulder, CO, USA).  Measurements were made at 1 nm intervals over the range 1250-2350 nm. Each sample was scanned once per day for three separate days, and each individual scan consisted of 24 single measurements. All of the scans from one sample were then averaged into one resulting spectrum. The spectral data were recorded as reflectance 4  values and it was transformed to log 1/R (where R is reflectance) before it was used by the calibration modelling software. Full cross validation procedure was used to validate our model because of the relatively small sample size (Martens & Dardenne, 1998; Terhoeven-Urselmans et al., 2006). Each set of spectra was divided into 12 segments and a calibration model was calculated for each segment by only leaving one sample out. Then the model calculated was used to predict the value of the left-out sample to see how close it was to its mass spectrometer value. This process was then repeated for all 12 segments and until every sample had been left out once. Calibrations were calculated by partial least-squares regression (PLSR) procedures using The Unscrambler (CAMO Software AS. Oslo, Norway). All the samples including the outliers were used for calibration and the related statistical analysis. The optimal number of Principal Components (PCs) used in the model was selected based on the least amount of PCs used while keeping the largest coefficient of multiple determinations for the calibration model (R2) and the lowest residual variance found. Standard error of prediction (SEP) is a measurement of the difference between the actual and predicted property values calculated over all cross validation calibrations (Kleinebecker et al., 2009). Standard error of calibration (SEC) is exclusively based on spectra used for calibration and indicates the theoretical accuracy when using the calibration to predict unknown spectra (Kleinebecker et al., 2009). The ratio of standard deviation to root mean standard error of the prediction (RPD) was used as a measurement for the model best fit.  Models with RPD larger than 2 are consider to have good predictions, and acceptable models should have a RPD value between 1.4 to 2 (Chang et al., 2001). Analysis was also done with spectra of each sample to try to isolate the section of wavelength most sensitive to the changes of the amount of δ13C[‰] and δ15N[‰] values. This was done by dividing the spectrum of the ambient CO2 treated clone to the spectrum of its enriched CO2 treated counterpart. An average of all the ambient CO2 treated spectra was also divided by the average of all the enriched CO2 treated spectra for this analysis.   5  Results The δ13C values used for this calibration model have a very wide variation of 19.49‰ and δ15N values have a very narrow range of 3.05‰ (Table 2). The RSD values for both the δ13C and δ15N calibration models were below the acceptable 1.4 limit to be used as adequate models (Table 1). The calibration models generated were unable to accurately predict δ13C and δ15N due to their high SEP values and low coefficients of multiple determinations for the cross validation models (r2) (Table 1, Fig 1 a-b). Four more separate calibration models were calculated using only ambient or enriched CO2 treated samples to predict δ13C and δ15N values. Even with the narrower range of the δ13C values used in the calibrations, the accuracy of the ambient models only improved slightly while the enriched model did not. All of those models had RPD values below or too close to the 1.4 limit (Table 1, Fig 1 c-f).   With the exceptions of samples from the Quesnel River clone, almost all ambient CO2 treated samples have higher reflectance values than their enriched counterparts (Fig 2 a). When I divided the ambient CO2 treated clone's spectrum over its enriched CO2 treated counterpart, the most dramatic change in reflectance for most samples occurred within the 2050 to 2150nm range. All of the samples from different clones have different change in reflectance patterns yet some samples from the same clones have similar changes in their reflectance patterns (Fig 2 b-g). Table 1: NIR calibration and cross validation statistics of PLSR models δ13C[‰]  Log 1/R  δ15N[‰]  Log 1/R   Complete Ambient  Enriched  Complete Ambient  Enriched No. Factors 6 5 1 No. Factors 6 9 2 R2 0.863 0.954 0.076 R2 0.856 0.999 0.549 SEC 4.221 0.234 0.569 SEC 0.446 0.029 0.776 r2 0.671 0.709 -0.690 r2 0.630 0.624 0.282 SEP 6.433 0.571 0.634 SEP 0.727 0.608 0.928 RMSEP 6.335 0.549 0.617 RMSEP 0.716 0.586 0.903 RPD 0.090 1.422 0.925 RPD 1.296 1.323 1.027 Outliers 2 1 0 Outliers 14 2 3  Transformations for regression analyses: log 1/R (R = reflectance), R2, coefficient of multiple determination; SEC, standard error of calibration; r2, coefficient of multiple determination for the cross validation model; SEP, standard error of prediction by cross validation; RMSEP, root mean square of the SEP; RPD, ratio of the standard deviation of the reference values to RMSEP.  6  Table 2: Statistical summary of the sample set      Sample CO2 Leaf  δ15N [‰] Leaf  δ13C [‰] I1BL Ambient -1.88 -24.48 I1BL Enriched -1.58 -40.25 I1CL Ambient -2.62 -24.18 I1CL Enriched -2.11 -41.03 I1LA Ambient -4.12 -24.76 I1LA Enriched -2.35 -40.54 I1LB Ambient -3.10 -23.39 I1LB Enriched -2.38 -40.17 I1LC Ambient -2.53 -22.72 I1LC Enriched -2.34 -41.65 J1BL Ambient -4.38 -24.33 J1BL Enriched -4.27 -41.61 J1LA Ambient -3.99 -25.35 J1LA Enriched -2.81 -40.82 J1LB Ambient -4.38 -24.33 J1LB Enriched -4.27 -41.61 J2AL Ambient --- --- J2AL Enriched -4.02 -42.20 J2LA Ambient --- --- J2LA Enriched -4.02 -42.20 J2LB Ambient --- --- J2LB Enriched -3.76 -41.46 J4AL Ambient -3.04 -24.85 J4AL Enriched -3.55 -41.24 J4BL Ambient -3.83 -25.41 J4BL Enriched -2.27 -40.92 J4LA Ambient -3.04 -24.85 J4LA Enriched -3.55 -41.24 J4LB Ambient -3.83 -25.41 J4LB Enriched -2.27 -40.92 Q2LB Ambient -3.32 -24.28 Q2LB Enriched -2.89 -41.14 Q3LA Ambient --- --- Q3LA Enriched -1.34 -41.45 Q3LB Ambient --- --- Q3LB Enriched -1.83 -40.52 Q4LA Ambient --- --- Q4LA Enriched -3.05 -41.26  Leaf  δ15N  Leaf  δ13C  For All Samples Number 32  Mean -3.09 -34.39 Standard Deviation 0.893 8.350 Range -4.38 -42.20  -1.34 -22.72 Ambient Samples Only Number 13  Mean -3.390 -24.487 Standard Deviation 0.775 0.781 Range -4.384 -25.412  -1.876 -22.716 Enriched Samples Only Number 19  Mean -2.877 -41.171 Standard Deviation 0.928 0.571 Range -4.270 -42.201  -1.339 -40.168 7   (a) Predicting δ13C [‰] using all the samples.   (b) Predicting δ15N [‰] using all the samples.  Figure 1: Relationship between NIRS predicted values and measurements by reference methods. For prediction, the partial least-square regression (PLSR) calibration model with the best cross validation statistics is presented. NIRS Predicted Values Actual Values NIRS Predicted Values Actual Values 8   (c) Predicting δ13C [‰] using ambient CO2 treated samples only.  (d) ) Predicting δ15N [‰] using ambient CO2 treated samples only.  NIRS Predicted Values Actual Values NIRS Predicted Values Actual Values 9   (e) Predicting δ13C [‰] using enriched CO2 treated samples only.  (f) Predicting δ15N [‰] using enriched CO2 treated samples only. NIRS Predicted Values Actual Values NIRS Predicted Values Actual Values 10   Figure 2: Ambient / Enriched CO2 treatment changes in reflectance spectra graphs  (a) Average ambient CO2 treated samples / average enriched CO2 treated samples 0.97 0.98 0.99 1 1.01 1.02 1.03 1.04 1250 1276 1302 1328 1354 1380 1406 1432 1458 1484 1510 1536 1562 1588 1614 1640 1666 1692 1718 1744 1770 1796 1822 1848 1874 1900 1926 1952 1978 2004 2030 2056 2082 2108 2134 2160 2186 2212 2238 2264 2290 2316 2342 Wavelenght (nm) Ambient / Enriched 11    0.94 0.96 0.98 1 1.02 1.04 1.06 1.08 1.1 1.12 1250 1298 1346 1394 1442 1490 1538 1586 1634 1682 1730 1778 1826 1874 1922 1970 2018 2066 2114 2162 2210 2258 2306 I1BL I1CL Wavelength (nm) Ambient / Enriched 0.94 0.96 0.98 1 1.02 1.04 1.06 1.08 1250 1298 1346 1394 1442 1490 1538 1586 1634 1682 1730 1778 1826 1874 1922 1970 2018 2066 2114 2162 2210 2258 2306 I1LA I1LB I1LC Wavelengeth (nm) Ambient / Enriched 0.94 0.96 0.98 1 1.02 1.04 1.06 1.08 1.1 1250 1298 1346 1394 1442 1490 1538 1586 1634 1682 1730 1778 1826 1874 1922 1970 2018 2066 2114 2162 2210 2258 2306 J1BL J1LA J1LB Wavelength (nm) Ambient / Enriched (b) Ambient / enriched CO2 treated samples from the Bell-Irving River clone replicates (I1) (d) Ambient / enriched CO2 treated samples from the Jasper River clone replicates (J1) (c) Ambient / enriched CO2 treated samples from Bell-Irving River clone replicates (I1L) 12      0.94 0.96 0.98 1 1.02 1.04 1.06 1.08 1250 1298 1346 1394 1442 1490 1538 1586 1634 1682 1730 1778 1826 1874 1922 1970 2018 2066 2114 2162 2210 2258 2306 J2AL J2LA J2LB Ambient / Enriched Wavelength (nm) 0.94 0.96 0.98 1 1.02 1.04 1.06 1.08 1.1 1250 1298 1346 1394 1442 1490 1538 1586 1634 1682 1730 1778 1826 1874 1922 1970 2018 2066 2114 2162 2210 2258 2306 J4AL J4BL J4LA J4LB Ambient / Enriched Wavelength (nm) 0.75 0.8 0.85 0.9 0.95 1 1.05 1250 1298 1346 1394 1442 1490 1538 1586 1634 1682 1730 1778 1826 1874 1922 1970 2018 2066 2114 2162 2210 2258 2306 Q2LB Q3LA Q3LB Q4LA Ambient / Enriched Wavelength (nm) (e) Ambient / enriched CO2 treated samples from the Jasper River clone replicates (J2) (g) Ambient / enriched CO2 treated samples from the Quesnel River clone replicates (f) Ambient / enriched CO2 treated samples from the Jasper River clone replicates (J4) 13  Discussion Although I was unable to find a calibration model that can accurately predict the δ13C and δ15N values in Populus trichocarpa leaves, the NIRS was able to differentiate samples that have been treated with enriched CO2 concentration from its ambient counterparts. This is evident as almost all ambient CO2 treated samples have higher reflectance value than its enriched counterpart (ambient / enriched > 1) (Fig 2 a).  I had also found that the calibration models with only ambient CO2 treatment samples outperformed the calibration models with only enriched samples at predicting δ13C and δ15N values (Table 1). This suggests that the CO2 treatment has a tremendous effect on the reflectance properties of the sample that are not dependent on the δ13C and δ15N values. This result was consistent with the findings reported by Clark et al. (1995) where they found the NIRS could correctly identify the δ13C value of alfalfa (Medicago sativa L.) and several cool-season perennial grass samples 77 to 82% of the time but only for the  δ13C values that were in the lower 20% of their dataset. The CO2 enriched treatment not only interfered with the δ13C calibration model but with the δ15N model as well. According to Chang et al. (2001), the influence of wavelength on the correlation between reflectance intensity and total carbon and nitrogen content were similar. They believed that the similarity in spectral response for carbon and nitrogen was due to their high intercorrelation.   Further study is needed to be done on the CO2 effect in order for us to have a more accurate prediction of δ13C and δ15N values in leaves with higher carbon content. An experiment with more than two CO2 concentration treatments and more in-depth analysis from the mass-spectrometer would help create calibration models with many different reference values. This may hopefully all us to understand what affects the reflectance intensity the most in leaves with high carbon content.   The NIRS not only picked up the effect of CO2 concentration on the samples, it also picked up the variables that were more dependent on the location of where the samples were originated. As shown in Figure 2 b to g, each samples from different clones reacted to the CO2 treatment differently but samples of the same clone replicates seem 14  to follow a distinct pattern. Chang et al. (2001) in their experiment on soil, also noted that there was a tendency for regional similarity in NIR reflectance spectra of soils. They noted that all the soil had peaks in their spectra at similar wavelengths but soil originating from different regions peaks at different intensity. Sun et al. (2012) were also able to classify lamb meat between pastoral regions from agricultural region using NIRS. They also had an 80% accuracy in classifying lamb meat samples from five different geographical origins.  Perhaps a better calibration model for predicting δ13C and δ15N values can be produced if both the effect of CO2 concentration and clonal variation can be minimize. More samples are needed from each population to minimize the effect of clonal variations. Clone replicates subject to CO2 with different isotopic compositions without changing the overall CO2 concentration can also minimize the effect of CO2 concentration. Through the above suggestions, the resulting calibration model should yield results similar to those that are done by Kleinebecker et al. (2009). This will then translate to more accurate predictions of δ13C and δ15N values with samples of various species at from different locations. Although there are no research performed on predicting δ13C and δ15N values from woody plant species to this day, a study done by Petisco et al. (2005) was performed by a NIRS to predict the nitrogen, phosphorus and calcium content in leaf samples in of several woody species from various locations. In that study, they were able to successfully predict the total nitrogen content in leaves, and they were also able to conclude that wavelengths between 2,024 and 2,176 nm to be the most relevant in their nitrogen calibration. They came to that conclusion by looking at the equations used by the calibration processes. At another study done on lamb meat, Sun et al. (2012) found success at predicting δ13C and δ15N values. They suggested the spectra ranged between 1380 and 1530 nm to be most relevant in their calibration model. This was consistent with the data of this study as the second highest change in the reflectance spectra of the average ambient over enriched CO2 treatment also occurs at the spectra ranged between 1380 and 1530 nm (Figure 2 a). Sun et al. (2012) theorized that the strong absorptions observed at the 1380 and 1530 nm relates to the C-H combination 15  vibrations and N-H first overtones. They believed that those bond vibrations were probably associated with carbohydrate and protein compounds in the defatted lamb meat. Based on the above findings, locating the wavelengths most relevant to δ13C and δ15N values for plants should be possible as long as there is an accurate model that can predict these values.  Neglecting the fact that NIRS was unable to perform isotopic analysis for woody plant species, NIRS is still a very powerful tool for plant physiology. The use for this technology are endless as NIRS is a proxy measurement that have many different correlations with many different elements. The limiting application of NIRS is in finding a suitable data pre-treatment and calibration strategies (Chang et al., 2001). One relatively quick and non-destructive scan of NIRS can estimate the total nitrogen, lignin, cellulose and many other different properties in a leaf of a woody plant from a sample set with several species coming from clearly different environmental conditions and leaf morphology and physiology (Petisco et al., 2006). Therefore NIRS has the potential to be used as a quick-screening and high-throughput phenotyping tool for leaf chemistry and many other plant physiology applications.  16  Reference  Bowling, D.R., D.E. Pataki & J.T. Randerson. 2008. Carbon isotopes in terrestrial ecosystem pools and CO2 fluxes. New Phytologist 178:24–40.  Buschhaus, H.A., Kalcsits, L.A. & Guy, R.D. 2007. ¹⁵N discrimination as an indicator of nitrogen dynamics in Populus trichocarpa. https://circle.ubc.ca/handle/2429/31914  Chang, C.W., D.A. Laird, M.J. Mausbach & C.R. Hurburgh. 2001. Near-infrared reflectance spectroscopy – Principal components regression analyses of soil properties. Soil Science Society of America Journal 65:480-490. Clark, D.H., D.A. Johnson, K.D. Kephart & N.A. Jackson. 1995. Near Infrared Reflectance Spectroscopy Estimation of 13 C Discrimination in Forages. Journal of Range Management 48:132–136. Clark, D.H., H.F. Mayland & R.C. Lamb. 1987. Mineral analysis of forages with near-infrared reflectance spectroscopy. Agronomy Journal 79:485–490. Dawson, T.E., S. Mambelli, A.H. Plamboeck, P.H. Templer & K.P. Tu. 2002. Stable isotopes in plant ecology. Annual Review of Ecology and Systematics 33:507–559. Farquhar, G.D., M.H. O’Leary & J.A. Berry. 1982. On the relationship between carbon isotope discrimination and the intercellular carbon dioxide concentration in leaves. Australian Journal of Plant Physiology 9:121-137 Kleinebecker, T., S.R. Schmidt, C. Fritz, A.J.P. Smolders & N. Holzel. 2009. Prediction of δ13C and δ15N in plant tissues with near-infrared reflectance spectroscopy. New Phytologist 184:732-739 Petisco, C., B. García-Criado, B.R. Vázquez de Aldana, I. Zabalgogeazcoa, S. Mediavilla & A. García-Ciudad. 2005. Use of near-infrared reflectance spectroscopy in predicting nitrogen, phosphorus and calcium contents in heterogeneous woody plant species. Analytical and Bioanalytical Chemistry 382:458-465 Petisco, C., B. García-Criado, S. Mediavilla, B.R. Vázquez de Aldana, I. Zabalgogeazcoa & A. García-Ciudad. 2006. Near-infrared reflectance spectroscopy as a fast and non-destructive toolto predict foliar organic constituects of several woody plant species. Analytical and Bioanalytical Chemistry 386:1823-1833 Martens, H.A., & P. Dardenne. 1998. Validation and verification of regression in small data sets. Chemometrics and Intelligent Laboratory Systems 44:99–121 17  Shenk, J.S., J.J. Workman & M.O. Westerhaus. 2008. Application of NIR spectroscopy to agricultural products. Pp 347-386 in Burns, D.A., & E.W. Ciurczak (ed.) Handbook of near-infrared analysis. CRC Press, Boca Raton, FL, USA Sun, Z.J., N.J. Livingston, R.D. Guy & G.J. Ethier. 1996. Stable carbon isotopes as indicators of increased water use efficiency and productivity in white spruce (Picea glauca (Moench) Voss). Plant, Cell and Environment 32:1821-1832 Sun, S., B. Guo, Y. Wei & M. Fan. 2012. Classification of geographical origins and prediction of δ13C and δ15N values of lamb meat by near infrared reflectance spectroscopy. Food Chemistry 135:508-514 Terhoeven-Urselmans, T., K. Michel, M. Helfrich, H. Flessa & B. Ludwig. 2006. Near-infrared spectroscopy can predict the composition of organic matter in soil and litter. Journal of Plant Nutrition and Soil Science 169:168–174 Workman, J.J. 2008. NIR spectroscopy calibration basics. Pp.123-150 in Burns, D.A., & E.W. Ciurczak (ed.) Handbook of near-infrared analysis. CRC Press, Boca Raton, FL, USA   18  Title: Evaluating the use of near-infrared reflectance spectroscopy as a proxy measure of carbon and nitrogen isotopes in the leaves of Black Cottonwood (Populus trichocarpa) Date: April 4th 2014 Name: Andy Leung Course number: Frst 498 Pimary Advisor: Professor Robert Guy Secondary Advisor: Professor Shawn Mansfield 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            data-media="{[{embed.selectedMedia}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.52966.1-0075640/manifest

Comment

Related Items