Sample preparation and measurement

In a greenhouse experiment we grew two tropical grass species, Digitaria eriantha and Urochloa mosambicensis, on soils with three different nitrogen treatments (0.05, 0.125 and 0.2 % N per 1 kg soil). The nitrogen levels in the soil were based on levels found in the Kruger National Park (KNP), South Africa, where both grass species occur as dominant species, and are considered to be important foraging grasses [van Oudtshoorn, 1992]. Pots were placed in a randomized layout. The conditions within the greenhouse were maintained at a constant temperature of 25◦C, with 12 daylight hours, and daily watering of the grasses. Three weeks after seed germination the number of grasses within the pots were thinned. To reduce soil background effects at the time of the spectral measurements, sufficient grass plants were

Chapter 2. Upscaling nutrient predictions

Table 2.1: Absorption features related to plant nitrogen [Cho and Skidmore, 2006; Ferwerda, 2005; Mutanga, 2004; Huang et al., 2004; Serrano et al., 2002; Martin and Aber, 1997; Curran et al., 1997; Fourty et al., 1996; Johnson et al., 1994; Matson et al., 1994; Curran et al., 1992; Card et al., 1988] λ (nm) Bond vibrations Biochemical

430 electron transition chl a 460 electron transition chl b 640 electron transition chl b 660 electron transition chl a 910a _{C − H stretch, 3}rd _{overtone protein}

1020a _{C − H stretch, 2}nd _{overtone protein}

1420 C − H stretch, C − H deformation protein, lignin 1510a _{N − H stretch, 1}st _{overtone protein, nitrogen}

1520a _protein

1690a _{C − H stretch, 1}st _{overtone protein, nitrogen, lignin,}

starch

1730a C − H stretch protein, cellulose, lignin 1940 O − H stretch, O − H deformation protein, nitrogen, lignin,

starch, water, cellulose 1950 O − H stretch, O − H deformation protein, nitrogen, lignin,

starch, water, cellulose 1960 O − H stretch, O − H bend protein, sugar, starch 1980a _{N = H asymmetry protein, lignin}

2060a _{N −H bend, N −H stretch, 2}nd_{overtone protein, nitrogen}

2130a _{N − H stretch protein}

2180a _{N −H bend, 2}nd_{overtone, C −H stretch,}

C − O stretch, C − N stretch

protein, nitrogen

2200a _protein

2240a _{C − H stretch protein}

2270 C −H stretch, O−H stretch, CH2bend,

CH2stretch

protein, nitrogen, lignin, starch, sugar,cellulose

2290a _protein

2300a _{N − H bend, C − O stretch, C − H bend,}

2nd_overtone

protein, nitrogen, cellu- lose

2350a _CH

2 bend, 2nd overtone, C − H defor-

mation, 2nd_overtone

protein, nitrogen, cellu- lose

REPda,b _nitrogen

REPca,b _nitrogen

REPga,b _nitrogen

a _{wavelength shown to relate to nitrogen in fresh material} b

red edge inflection point (REP), calculated using a standard derivative (REPd), linear four-point interpolation (REPg) and linear extrapolation (REPc), formulae in Cho and Skidmore [2006]

2.2. Methods

Table 2.2: Phenological classes and numbers of samples included in each class of spectral measurements Species Phenological stage Code Weeks from sowing Samples

D. eriantha Seedling DES 5 31∗

D. eriantha Adult DEA 8 38

U. mosambicensis Adult UMA 10 27

Spp. combined Adult & Seedling All NA 86

∗

Only 21 samples were included on the fresh material measurements

kept within each pot to ensure that, at the different phenological stages, the soil would be covered. As the germination rates of U.mosambicensis were low, there were only sufficient numbers of germinated seed to be included in a single phenological age, it was decided to measure them at the adult stage, at this phenological stage the plant material was sufficient to cover the soil.

Spectra were measured at two phenological stages: seedling and adult. At the seedling stage, morphologically the D.eriantha seedling canopy was a tufted, erect grass standing approximately 15–25 cm high. The adult stages of both D.eriantha and U.mosambicensis were without flower and reached a height of approximately 70 cm, U.mosambicensis formed denser tufts and their leaves were slightly broader than those of the adult D.eriantha plants. The number of spectral samples take for each of the phenological stages is presented in table 2.2. Spectra were measured using a GER 3700 spectroradiometer (Geophysical and Environmental Research Corp.). The GER 3700 is a three dispersion grating spectroradiometer using Si and PbS detectors with a single field of view. The instrument has a wavelength range of 350–2500 nm, the spectral range has a resolution of 1.5 nm in the 350–1050 nm range, 6.2 nm in the 1050–1900 nm range and 8.5 nm in the 1900–2500 nm range.

For fresh canopy material (referred to as ‘fresh material’ hereafter) mea- surements, the GER 3700 was fitted with a 10◦ optic, and placed on a tripod at a 15◦ angle, 80 cm above the pot rim level, creating a field of view (FOV) of 14 cm in diameter (the diameter of the pot size at rim level was 19 cm). Mounting the spectrometer at this height, above the pot rim, allowed for the projected plant growth between a seedling and an adult

Chapter 2. Upscaling nutrient predictions

growth phase, and ensured that the same setup could be maintained for all phenological stages. The calibration panel (Labsphere, Inc, Sutton, NH), used for converting the relative reflectance to absolute reflectance, was mea- sured at a distance of 31 cm from the optic, thus creating a FOV of 5.5 cm in diameter. The calibration panel was only 225 cm2, and thus a FOV of this size ensured an accurate reading of the panel. A halogen lamp was placed alongside the GER 3700 at the same level as the optic.

Potted plants were transferred from the greenhouse to the laboratory for measurement. The pot was fixed in place (to ensure the FOV was located above the pot centre) for a group reading, one group reading consisted of a measure of the calibration panel and a set of 5 readings of the sample. In order to reduce directional effects, caused by leaf orientation within the canopy, the pot was rotated by 90◦, and the next group reading taken [Cho and Skidmore, 2006; Mutanga et al., 2003]. The 20 spectral readings taken per pot were averaged to obtain a single spectral reading per sample. Following the fresh canopy material spectral measurement, all the grasses within a pot were clipped at the plant base and dried at 70◦C, for 24 hours. The samples were then ground through a 1 mm sieve (these samples are hereafter referred to as ‘dried material’). Spectra were measured with the GER 3700, fitted with a 3◦ optic, 38 cm above the point of measurement, creating a FOV of 2 cm. To reduce reflectance and directional effects, the dried material was placed in a shallow non-reflective bowl on a non- reflective background and the sample leveled. A measurement of a sample was composed of a calibration panel measurement and 5 readings of the sample. The 5 readings were averaged to create a single reading per sample. After completion of all spectral measurements, all the samples were chem- ically analysed. Nitrogen was analysed using the wet chemistry Kjedahl technique [AOAC, 1970].