265 
Cellulose (Cell. W), often referred as "crude fibre", was also determined using the method of Weende. 
Estimates of "Cell W" and "Cell S" differed significantly. This is probably due to the fact that "Cell W" and 
"Cell S" do not represent exactly the same chemical. Moreover, it is well known that estimates of 
concentrations from different chemical extraction methods tend to be different (Curran, 1989). 
N 
L 
Cell. W 
Cell. S 
Age 
N 
100 
90 
53 
22 
-71 
L 
90 
100 
56 
14 
-36 
Cell. W. 
53 
56 
100 
47 
-7 
Cell. S. 
22 
14 
47 
100 
4 
Range 
4.34 - 8.23 
16.3-22.5 
20.6 - 32.9 
21.7-30.6 
0-50 years 
Table 1 : Correlations between pine needle protein (N), lignin (L) and cellulose (CellW and CellS) 
concentrations. Ranges are expressed as a percentage of dry matter. The last column shows correlations 
between foliar chemical concentrations and the age of pine trees. 
* Spectral measurements 
All vegetation samples collected in the field were dried and ground down to similar particle sizes for further 
spectral analysis. The objective was to determine the most significant wavelengths for assessing chemical 
compounds. Total hemispherical spectral reflectances were recorded with a computerised spectrophotometer 
(Technicon InfraAnalyser 450) operated in a single-beam mode. This instrument measures spectral 
reflectances at 19 different wavelengths with the help of filters. Associated target compounds are indicated 
between brackets: 1445nm (wetness), 1680nm (reference), 1722nm (starch, cellulose, lignin), 1734nm 
(protein, cellulose), 1759nm (oil), 1778nm (cellulose, starch, fibres), 1818nm (cellulose), 1940nm (wetness), 
1982nm (protein), 2100nm (starch, cellulose), 2139nm (protein, starch), 2180nm (protein), 2190nm (protein, 
starch), 2208nm (urea), 2230nm (reference), 2270nm (starch, cellulose, lignin), 2310nm (oil), 2336nm 
(fibres, cellulose, lignin), 2348nm (cellulose, nitrogen). 
Figure 4 : Laboratory-derived foliar reflectance spectra, (a) InfraAlayser and (b) NIR 6500 instruments. 
Correlations were determined between laboratory-derived spectral reflectances and protein, lignin and 
cellulose concentrations. Two reflectance ratios (p/p 16 g 0 , P/P 2230 ) were computed with 2 reference 
reflectances (p 16g0 , p 2 23 o)- These ratios were meant for further comparison between laboratory- and remotely 
acquired results. Consequently, spectral bands within atmospheric absorption bands (i.e. P 1445 , p 1818 , P 1940 » 
p 2208 > P 2336 ’ P 2348 ) were discarded. Most correlations were larger than 50%, except for "Cell S" which yielded 
correlations that are much lower than those of "Cell W". In many cases, correlations tended to have similar 
values. This was explained by the intercorrelation of chemicals (Table 1), and not in terms of chemical bonds. 
This indicates the impossibility to have straightforward and unambiguous selections of predictive wavelengths.