Surface reflectance properties of Antarctic moss and their relationship to plant species, pigment composition and photosynthetic function


  • C. E. Lovelock,

    Corresponding author
    1. Smithsonian Environmental Research Center, PO Box 28, Edgewater MD 21037, USA and
      Dr Catherine Lovelock. Fax: +1 443 4822380; e-mail:
    Search for more papers by this author
  • S. A. Robinson

    1. Institute for Conservation Biology, University of Wollongong, Northfields Avenue, Wollongong, NSW 2522, Australia
    Search for more papers by this author

Dr Catherine Lovelock. Fax: +1 443 4822380; e-mail:


In this study the variations in surface reflectance properties and pigment concentrations of Antarctic moss over species, sites, microtopography and with water content were investigated. It was found that species had significantly different surface reflectance properties, particularly in the region of the red edge (approximately 700 nm), but this did not correlate strongly with pigment concentrations. Surface reflectance of moss also varied in the visible region and in the characteristics of the red edge over different sites. Reflectance parameters, such as the photochemical reflectance index (PRI) and cold hard band were useful discriminators of site, microtopographic position and water content. The PRI was correlated both with the concentrations of active xanthophyll-cycle pigments and the photosynthetic light use efficiency, Fv/Fm, measured using chlorophyll fluorescence. Water content of moss strongly influenced the amplitude and position of the red-edge as well as the PRI, and may be responsible for observed differences in reflectance properties for different species and sites. All moss showed sustained high levels of photoprotective xanthophyll pigments, especially at exposed sites, indicating moss is experiencing continual high levels of photochemical stress.


Solar radiation is essential for photosynthesis and growth of plants. The surface reflectance characteristics of leaves is key in determining both the quantity and quality of solar radiation that is incident on leaf internal organs (Vogelmann 1993). Although solar radiation in the visible region drives photosynthesis, absorptance of high levels of visible radiation and radiation at other wavelengths can be damaging. At shorter wavelengths, absorbed UV-B (280–320 nm) radiation can cause lesions to nucleic acid and proteins. Excess levels of visible radiation (400–750 nm) can cause photo-inhibition of photosynthesis (Cornic, Woo & Osmond 1982). High absorptance at longer wavelengths (> 750 nm) leads to increases in temperature that can be detrimental in arid environments where it is important for plants to conserve water by limiting transpiration and to maintain leaf temperatures close to that of air temperatures (Gates et al. 1965; Ehleringer 1981). Variation in absorptance of leaves has been observed (Ehleringer & Björkman 1978; Vogelmann 1993) and specialized adaptation of surface reflectance properties in plants is known to facilitate photosynthetic carbon gain. For example, in desert ecosystems pubescent or waxy leaf surfaces that increase reflectance are common (Billings & Morris 1951; Ehleringer 1981) and have been demonstrated to be important in protection from photo-inhibition of photosynthesis (e.g. Robinson, Lovelock & Osmond 1993). Moreover, epidermal tissues often accumulate compounds that are effective screens against UV-B radiation, thereby decreasing their reflectance in the UV region (e.g. Tevini 1993).

In this study we surveyed the natural variability of surface reflectance in moss in continental Antarctica. At the study site in Wilkes Land, eastern Antarctica, conditions are cold and usually very arid, resulting in lichens dominating the rocky, ice-free areas. The moss community is well developed for the region, but it is restricted to melt lakes, melt streams and other locations where free water is available for at least some of the summer (November–March) (Melick, Hovendon & Seppelt 1997). In winter the moss is covered with snow and ice and is subject to disturbance from cryo-perturbation, which causes highly complex microrelief of ridges and valleys within the moss turves. In comparison with lower latitudes, levels of UV radiation are low in Antarctica. But, Antarctica is experiencing large increases in incident UV-B radiation due to reductions in concentrations of stratospheric ozone (Frederick & Snell 1988). Increases in UV radiation are having large effects on the primary production of the oceans (Neale, Davis & Cullen 1998), and could also be affecting terrestrial vegetation. The heat balance of moss, photosynthetic processes and the impacts of enhanced UV radiation are likely to be partially dependent on moss surface reflectance characteristics and pigment composition. Thus, the first aim of this study   was   to   test   the   hypothesis   that   moss   species   and moss inhabiting different environments have different surface  reflectance characteristics. We measured the surface reflectance, at approximately 1 nm intervals from 200 to 900 nm, of three dominant moss species over three sites with differing water availabilities, and over differences in microtopography.

Surface reflectance properties of vegetation are often used as indicators of photosynthetic function in remote sensing of vegetation (reviewed in Field, Gamon & Penuelas 1994). Variation in reflectance has been observed among plant taxa (e.g. Gates et al. 1965; Peñuelas et al. 1993; Gamon et al. 1995), due to differences in water and nutrient availability (e.g. Peñuelas et al. 1994; Gamon, Serrano & Surfus 1997), environmental stress (e.g. Rock, Hoshizaki & Millar 1988), and both seasonally (Running & Nemani 1988; Gamon et al. 1995) and diurnally (Gamon et al. 1990; Peñuelas et al. 1994; Gamon et al. 1997). Changes in surface reflectance of intact leaves have been directly correlated with leaf chemical composition (Jacquemoud et al. 1996) and changes in photosynthetic processes in some higher plant taxa (Gamon et al. 1990, 1997; Gilmore & Ball 2000). However, it is often difficult to reconcile physiological measurements, which are usually made at the scale of individual leaves and remotely sensed measurements, which are often made at the scale of whole canopies (Williams 1991; Field et al. 1994). Some of the difficulty arises because canopies are complex. Canopies are composed of layers of leaves of different species that can have a range of properties. Leaves within canopies can have varying distributions, orientations, morphologies, internal structures, pigment compositions, and epidermal characteristics, all properties that could influence the surface reflectance of the canopy (Rock et al. 1988; Field et al. 1994; Vose et al. 1995). At the leaf level, mosses are very simple consisting of a few cell layers. But at a higher level of organization, moss turves, although small in comparison to the canopies of trees, are relatively complex, composed of many leafy gametophytes packed together to form what can be thought of as a ‘microcanopy’. In mosses, measurement of photosynthetic processes and surface reflectance can be made at the same scale. Thus, the second aim of our study was to test the hypothesis that the spectral properties of moss are directly related to photosynthetic processes and pigment composition. We wished to understand whether non-destructive hyperspectral measurements could potentially be used to map distributions of moss species and physiological activity both spatially and temporally. Herein we describe tests of correlations among surface reflectance properties and commonly used reflectance indices and pigment concentrations, water contents and photosynthetic function of Antarctic moss.

Materials and methods

Moss was obtained from three sites around the Australian base Casey, Wilkes Land, continental Antarctica (66°17′ S, 110°32′ E). Two sites were directly adjacent to the base (Red Shed and Science), whereas the other site was 20 km west at Robinson Ridge. Water availability varies across the sites. The Red Shed site is directly adjacent to a large summer melt lake, the Robinson Ridge site is adjacent to a melt stream that flows intermittently over the summer months, whereas at the Science site water is only available when overlying snow melts early in the summer. Temperature also varies over the sites. Robinson Ridge is cooler and more exposed, than the Red Shed and Science sites (Melick et al. 1997). In the early 1990s the Science site was exposed to contamination due to deposition of cement dust during construction (Adamson, Adamson & Seppelt 1994).

At each site representative samples, approximately 3 cm2 were cut from the turves over three summer seasons (October to February) from 1996 to 1999. The three most common species, Grimmia antarctici Card., Ceratodon purpureus (Hedw.) Brid., and Bryum pseudotriquetrum (Hedw.) Gaertn., Meyer and Scherb. were sampled. The necessity of minimizing the impacts of the study, the extent of the moss beds (smaller at Science) and the natural distribution of the three moss species resulted in different representations of each species at each site (Table 1). We tested for differences between moss species and the influence of site on pigment and surface reflectance parameters using samples from all sites. We used a subset of the Robinson Ridge and Red Shed Grimmia antarctici samples to test for effects of microtopography.

Table 1.  Details of species and site for moss samples used in the study
SpeciesNumber of samples per site
Red ShedScience
Ceratodon purpureus 4 45
Bryum pseudotriquetrum 7 32
Grimmia antarctici24362

In addition we tested the influence of moss water content on surface reflectance. We used four samples of G. antarctici from microtopographic ridges at Robinson Ridge. We measured the surface reflectance of the samples when they were dry, and after they had been rehydrated with a fine mist of water for 5 min. Excess water was blotted from the moss before measurements. Moss re-hydrate rapidly, regaining full physiological activity in 1 to 2 min (Robinson et al. 2000).

Measuring surface reflectance

We used an integrating sphere fitted to a scanning spectrophotometer (sphere diameter 63 mm; sample aperture size 3 cm2; GBC UV-Vis 918; GBC, Melbourne, Australia) to measure spectral reflectance of moss between 200 and 900 nm at approximately 1 nm intervals. The moss turf samples were large enough to completely fill the sample aperture of the integrating sphere. To maintain the moss canopy architecture (gametophyte arrangement) the samples were held in place within a 1 cm deep highly reflective sample holder. For some samples this required shaving tissue from the basal portion of the sample using a razor blade until the turf was 1 cm thick. The depth and packing of the samples resulted in transmittance through the sample being negligible. Reflectance measured therefore represents the moss canopy with green photosynthetic tissue of between 2 and 5 mm depth and 8–5 mm of non-photosynthetic tissue below. From the entire reflectance spectra, using subsets of the total 87 spectra (Table 1), we present mean reflectance at key wavelengths and also calculate reported indices used in remote sensing vegetation (Table 2). The cold hard band (CHB; Gilmore & Ball 2000) has been found to correlate with the formation of a chlorophyll-protein complex in leaves that protects against freezing damage. The photosynthetic reflectance index (PRI; Gamon et al. 1990, 1997) was developed to reflect changes in concentrations of the xanthophyll cycle pigments that are formed when plants are stressed. The amplitude of the reflectance change at the red-edge (δRE) and the position of the red-edge (δλRE) were calculated from the first derivative of the spectra (Horler, Dockray & Barber 1983). Other reflectance indices used by a range of researchers were also calculated (e.g. Carter 1991, 1993; Vogelmann, Rock & Moss 1993; Lichtenthaler, Gitelson & Lang 1996; Gitelson & Merzlyak 1997).

Table 2.  Calculation of reflectance indices
  1. aGilmore & Ball 2000; bPeñuelas et al. 1994; cRouse et al. 1973; dGamon et al. 1997.

Cold hard bandaCHB(R750 − R710)/(R750 + R710)
Normalized difference between the first derivative in the red and greenbEGFN(1st R530 − 1st R710)/(1st R530 − 1st R710)
Ratio of first derivative in the green to first derivative in the redbEGFR(1st R530/1st R710)
Normalized difference vegetation indexcNDVI(R850 − R680)/(R850 + R680)
Normalized pigment chlorophyll ratio indexbNPCI(R680 − R430)/(R680 + R430)
Photosynthetic reflectance indexdPRI(R531 − R570)/(R531 + R570)

Pigment, water and photosynthetic characteristics

Prior to measurement of the reflectance spectra, a subset of the G. antarctici samples were dark-adapted for 20 min after which the chlorophyll fluorescence parameter Fv/Fm was measured using a PAM 2000 (H. Walz, Effeltrech, Germany). After the reflectance spectra were measured, the photosynthetically active apices of the moss were removed using a razor blade and the tissue was frozen in liquid nitrogen. Samples were returned to Australia in liquid nitrogen and then stored at −80 °C in a freezer prior to pigment analysis. For chlorophyll and carotenoid determination, samples [50–100 mg fresh weight (FW)] were ground with liquid nitrogen and sand in a mortar and pestle and then extracted in (1·5 mL) 100% acetone. The samples were transferred to a microcentrifuge tube containing 1 mg sodium bicarbonate and kept on ice in the dark for 20 min After centrifugation (13 600 × g for 5 min) the pellet was re-extracted in 0·5 mL 80% acetone using a polypropylene tissue grinder (Crown Scientific, Sydney, Australia). After a further 20 min on ice, the sample was centrifuged as above and the supernatants from each extraction combined and made up to 3 mL with 100% acetone. Chlorophylls and carotenoids were then quantified by high-pressure liquid chromatography (HPLC) using a method adapted from Gilmore & Yamamoto (1991). The samples (30–70 µL) were injected into a Shimadzu HPLC system (Shimadzu, Sydney, Australia) at a flow rate of 2 mL min−1. Solvent A (acetonitrile : methanol : Tris HCL buffer 0·1 m, pH 8·0; 79 : 8 : 3) ran isocratically from 0 to 4 min followed by a 2·5 min linear gradient to 100% Solvent B (methanol : hexane; 4 : 1) which then ran isocratically from 6·5 to 15 min Flow rate was decreased from 2 to 1·5 mL min−1 between 6 and 12 min and then ran at 1·5 mL min−1 until 15 min, to maintain stable pressure. The column was re-equilibrated with solvent A between samples. Pigments were separated on an Spherisorb ODS1 column (Alltech, Sydney, Australia) and quantified by integration of peak areas, detected at 440 nm using a photodiode array detector (Model SPD-M10AVP; Shimadzu), relative to pure chlorophyll (Sigma, Sydney, Australia) and carotenoid (Extrasynthase, Genay, France and VKI, ­Horsholm, Denmark) standards.

Anthocyanin concentrations were determined using the differential pH method of Fuleki & Francis (1968) and Francis (1982). Moss tissue (0·1–0·2 g FW) was ground with liquid nitrogen and sand in a mortar and pestle and then extracted in (1·5 mL) of 1% HCl in methanol. Samples were centrifuged at 13 600 × g for 6 min and absorbance of the supernatant measured at 526 nm One hundred micro­litres of sodium acetate buffer (pH 5·0) was then added to 1 mL of the supernatant and after 3 min this sample was centrifuged for 10 min at 13 600 × g before a second measurement of absorbance at 526 nm. Anthocyanin ­concentration was calculated from the difference in these two absorbance measures.

Water content of samples was measured after oven drying at 80 °C to stable weight and is expressed as g H2O g [dry weight (DW)]−1 as described in Robinson et al. (2000). Concentration of UV-B absorbing pigments was determined from 15 to 60 mg of dry moss tips. Dry tissue was weighed, ground in liquid nitrogen with acid-washed sand and then extracted in 1·5 mL of acidified methanol (methanol–H2O–HCl; 79 : 20 : 1). The sample was transferred to a microfuge tube and allowed to stand on ice in the dark, with vortexing every 20 min After 1 h the tube was centrifuged (13 600 × g, 4 min) and the supernatant removed. The final volume of supernatant was made up to 1·5 mL with acidified methanol. Absorbance across the UV-B region (280–320 nm) was scanned spectrophotometrically. Values are expressed as mean absorbance for this wavelength range.

Data analysis

Correlations among surface reflectance indices, pigments, chlorophyll fluorescence parameter Fv/Fm and water content were tested using the non-parametric Spearman rank correlation (Rho) using the statistical computing package Data Desk 6·1 (Version 6·1; Data Descriptions, Ithaca, NY, USA). Rho detects consistently increasing or decreasing trends but does not assume variables are linearly related. Linear regression was also used to test for significant relationships between moss chlorophyll concentrations and reflectance indices, and among other variables. Tests of the influence of site, species and microtopography on pigments and reflectance parameters were performed using analysis of variance (anova). For tests of sites and species anovas were unbalanced (unequal sample sizes), therefore post hoc tests of differences between individual means were adjusted for the unequal sample sizes using the Bonferroni test. The adequacy of the anova models was assessed by inspecting residual plots. Paired t-tests were used to assess the influence of hydration on reflectance parameters.


Differences between species in reflectance and pigments

Reflectance spectra of mosses were similar in shape to those observed for intact leaves of angiosperms with the major features of the spectra those indicative of chlorophyll (Fig. 1). Examination of the first derivative of the spectra (Fig. 1b) showed typical peaks at the green (approximately 525 nm) and red edge (approximately 700 nm). Moss species did not significantly differ in their UV reflectance (R320), but they differed significantly at 526, 550 and 850 nm, in both δRE and δλRE, and tended to have different CHB (Table 3). In all cases, B. pseudotriquetrum had greater reflectance values.

Figure 1.

Mean reflectance spectra (a) and first derivative (b) of all samples of the three common Antarctic moss species: Bryum pseudotriquetrum (dotted line, n = 3), Ceratodon purpureus (dashed line, n = 5) and Grimmia antarctici (solid line, n = 25) collected during the 1997–98 season from the Robinson Ridge and Red shed sites.

Table 3.  Surface reflectance characteristics and pigment concentrations for three common moss species in Continental Antarctica, (n=11, 12, 63, respectively)
 Bryum pseudotriquetrumCeratodon purpureusGrimmia antarcticiFP
  1. NS signifies a non-significant difference at the P=0·10 level.

 R320 0·0070 ± 0·0008 0·0071 ± 0·0010 0·0070 ± 0·0003NS 
 R526  0·049 ± 0·005  0·022 ± 0·006  0·023 ± 0·00128·40< 0·0001
 R550  0·055 ± 0·007  0·034 ± 0·008  0·034 ± 0·002 5·840·0074
 R850  0·388 ± 0·037  0·275 ± 0·020  0·311 ± 0·012 4·290·0170
 δRE 0·0101 ± 0·0001 0·0060 ± 0·0001 0·0068 ± 0·0001 8·740·0004
 δλRE  705·4 ± 1·9  700·6 ± 1·0  700·4 ± 0·5 7·330·0012
 PRI−0·160 ± 0·028−0·160 ± 0·015−0·172 ± 0·009NS 
 NDVI  0·835 ± 0·047  0·800 ± 0·029  0·868 ± 0·012NS 
 NPCI  0·442 ± 0·057  0·484 ± 0·043  0·398 ± 0·024  
 EGFR   6·04 ± 1·58   8·82 ± 2·43   8·03 ± 0·85NS 
 EGFN  0·643 ± 0·042  0·784 ± 0·072  0·701 ± 0·021NS 
 CHB  0·349 ± 0·058  0·229 ± 0·021  0·261 ± 0·015 2·740·0708
 Anthocyanins (A526 diff g−1 FW)   1·2 ± 0·2    1·3 ± 0·2    1·2 ± 0·2NS 
 Anthocyanins/TChl. (A526 diff mol−1)   2·9 ± 0·3    5·7 ± 1·2    3·0 ± 0·2 7·420·0011
 UV absorbing pigments   Mean A(320−280nm) (g−1 DW)      266 ± 42      127 ± 22      110 ± 915·34< 0·0001
 Total chlorophyll (nmol g−1 FW)      475 ± 60      241 ± 40      453 ± 44 2·500·0882
 Chl a : b ratio   3·2 ± 0·2    3·6 ± 0·3    3·2 ± 0·1NS 
 VAZ/TChl (mmol mol−1)      64 ± 5      132 ± 36      103 ± 6 3·640·0305
 %(AZ/VAZ)   18·0 ± 3·9   39·4 ± 4·9   37·4 ± 2·5 6·370·0026
 Tcar./Tchl. (mmol mol−1)      443 ± 15      643 ± 96      631 ± 31 2·860·0629

Pigment concentrations also varied among species (Table 3). Bryum pseudotriquetrum had higher levels of UV-absorbing   pigments   but   lower   carotenoid   levels   (total carotenoid and xanthophyll concentrations) in comparison with C. purpureus and G. antarctici. Ceratodon purpureus  had lower total chlorophyll concentrations than G. antarctici or B. pseudotriquetrum but higher levels of anthocyanins.

Many of the moss pigments concentrations tended to co-vary with each other. For example, total chlorophyll concentration was highly correlated with the total carotenoid content (Rho = 0·91). Additionally, concentrations of all the carotenoids tended to be highly correlated (e.g. correlation among lutein and the xanthophyll cycle pigments Rho = 0·726). Anthocyanins, measured as absorptance of extracts at 526 nm and expressed on a fresh weight basis, were also correlated with total chlorophyll (Rho = 0·625). The proportion of xanthophyll-cycle pigments in the de-epoxidated photoprotective forms, zeaxanthin and antheraxanthin, was negatively correlated with total chlorophyll (Rho = −0·482) and positively correlated with the total pool size of the xanthophyll-cycle pigments (Rho = 0·588).


The PRI was strongly influenced by whether moss was growing on a ridge or valley, but no other reflectance parameter was significantly affected by microtopographic position (Table 4). Samples from ridges had lower values of PRI (more negative) than those in the valleys. Total chlorophyll concentrations were approximately half on the ridges compared to the valleys, whereas the concentration of xanthophyll-cycle pigments per chlorophyll was 50% higher. There was also a significantly higher proportion of the xanthophyll-cycle present in the photoprotective forms on ridges.

Table 4.  Surface reflectance properties and pigment concentrations of Grimmia antarctici over different microtopographic positions. Moss is from ridges or valleys at Robinson Ridge and Red Shed sites, n = 8
  1. NS signifies a non-significant difference at the P=0·10 level.

 R320 0·0058 ± 0·0003 0·0066 ± 0·0006NS 
 R526  0·027 ± 0·003 0·0022 ± 0·001NS 
 R550  0·035 ± 0·004  0·042 ± 0·003NS 
 R850  0·362 ± 0·016  0·374 ± 0·023NS 
 δRE 0·0079 ± 0·0001 0·0084 ± 0·0001NS 
 δλRE 700·75 ± 1·26700·25 ± 1·19NS 
 PRI−0·213 ± 0·009−0·117 ± 0·01148·790·0001
 NDVI  0·849 ± 0·013  0·882 ± 0·031NS 
 NPCI  0·333 ± 0·064  0·425 ± 0·069NS 
 EGFR   8·90 ± 1·77   9·33 ± 2·16NS 
 EGFN  0·761 ± 0·032  0·780 ± 0·043NS 
 CHB  0·295 ± 0·025  0·256 ± 0·036NS 
 Anthocyanins (A526 diff g−1 FW)   1·0 ± 0·4   1·9 ± 0·3NS 
 Anthocyanins/TChl. (A526 diff mol−1)   2·7 ± 0·5   2·6 ± 0·6NS 
 Total chlorophyll (nmol g−1 FW)     375 ± 73     791 ± 6617·880·0008
 Chl a : b ratio   3·3 ± 0·2   3·1 ± 0·2NS 
 VAZ/TChl (mmol mol−1)     130 ± 12      81 ± 612·870·0030
 %(AZ/VAZ)   40·9 ± 5·4   27·1 ± 3·3 4·820·0454
 Tcar./Tchl. (mmol mol−1)     761 ± 87     465 ± 1810·980·0051


Over the three sites we observed significant difference in many of the reflectance parameters, but there were no significant effects of site on R320, PRI, NDVI or NPCI (Table 5). The Science site had higher reflectance at 526 and 550 nm, but lower reflectance at 850 nm in comparison with moss growing at the Red Shed and Robinson Ridge sites. Moss at the Red Shed site exhibited greater δRE and δλRE values than the moss at the Robinson Ridge and Science sites. The CHB was also higher at Robinson Ridge than at the other two sites. Moss at the Red Shed site had lower EGFR and EGFN than moss at Robinson Ridge and Science.

Table 5.  Surface reflectance characteristics and pigment concentrations of moss over three sites in Continental Antarctica. n = 34, 43, 9, respectively
 Robinson RidgeRed ShedScienceFP
  1. NS signifies a non-significant difference at the P=0·10 level.

 R3200·0068 ± 0·00040·0070 ± 0·00040·0086 ± 0·0011NS 
 R5260·0027 ± 0·00040·0023 ± 0·00040·0038 ± 0·0011 4·940·0094
 R550 0·035 ± 0·002 0·034 ± 0·001 0·052 ± 0·004 3·430·0371
 R850 0·351 ± 0·017 0·289 ± 0·015 0·331 ± 0·024 4·280·0170
 δRE0·0081 ± 0·00010·0063 ± 0·00010·0070 ± 0·0001 3·930·0235
 δλRE 703·8 ± 1·1699·3 ± 0·2 700·1 ± 0·411·91< 0·0001
 PRI−0·165 ± 0·014−0·177 ± 0·010−0·146 ± 0·011NS 
 NDVI 0·857 ± 0·019 0·864 ± 0·015 0·808 ± 0·042NS 
 NPCI 0·454 ± 0·035 0·391 ± 0·027 0·370 ± 0·051NS 
 EGFR 10·19 ± 1·43 5·86 ± 0·58 9·30 ± 2·99 4·620·0127
 EGFN 0·764 ± 0·032 0·651 ± 0·023 0·742 ± 0·062 4·360·0159
 CHB 0·324 ± 0·029 0·226 ± 0·008 0·256 ± 0·026 4·520·0137
 Anthocyanins (A526 diff g−1 FW)  1·5 ± 0·2  1·1 ± 0·2  1·3 ± 0·1NS 
 Anthocyanins/TChl. (A526 diff mol−1)  3·5 ± 0·5  3·3 ± 0·4  3·3 ± 0·3NS 
 UV absorbing pigments Mean A(320−280nm) (g−1 DW)    213 ± 37    132 ± 20    104 ± 17 4·290·0224
 Total chlorophyll (nmol g−1 FW)    500 ± 58    374 ± 49    393 ± 39NS 
 Chl a : b ratio  3·2 ± 0·2  3·3 ± 0. 1  3·4 ± 0·2NS 
 VAZ/TChl (mmol mol−1)    124 ± 14     93 ± 6     61 ± 7 4·720·0114
 %(AZ/VAZ)  29·7 ± 3·0  40·4 ± 3·0  28·5 ± 6·3 3·680·0292
 Tcar./Tchl. (mmol mol−1)    677 ± 55    586 ± 29    458 ± 16 3·190·0462

Differences in reflectance indices corresponded to changes in pigment contents over the sites. Total chlorophyll and anthocyanin concentrations were not significantly different among sites, although the mosses tended to have higher concentrations of these pigments at Robinson Ridge than at the Science and Red Shed sites (Table 5). Xanthophyll pigment concentrations, total carotenoids and concentrations of UV absorbing pigments were all greater at Robinson Ridge than at the other two sites.

Relationships among reflectance, plant pigments, water content and photosynthesis

Overall, correlation between surface reflectance indices and plant pigment contents was low. Concentration of total chlorophyll was correlated significantly, but not strongly with many of the key reflectance wavelengths and indices (Table 6). The strongest linear regression was between total chlorophyll and reflectance indices calculated with reflectance measured at 850 and 680 nm (either R850, R850–680, and NDVI) and with δRE (Fig. 2). There was no significant relationship between total chlorophyll and EGFR, EGFN or NPCI.

Table 6.  Reflectance signals of moss over all sites and species, and their Spearman rank correlation (Rho) and linear relationship with chlorophyll concentrations. Reflectance signals are from key wavelengths over the spectra measured. Regressions were carried out on logarithmically transformed data where appropriate
Reflectance parameterRhoRegressionR2
  1. Levels of significance are *P < 0·10, **P < 0·05, ***P < 0·001.

R5260·298LogR526 = −1·72 + 0·000193 × TChl0·060**
R5500·362LogR550 = −1·60 + 0·000242 × TChl0·080***
R8500·486R850 = 0·261 + 0·000134 × TChl0·166***
R850-2800·484R850-280 = 0·254 + 0·000133 × TChl0·161***
R850-3200·482R850-320 = 0·253 + 0·000133 × TChl0·163***
R850-4800·485R850-480 = 0·251 + 0·000134 × TChl0·166***
R850-5260·475R850-526 = 0·239 + 0·000125 × TChl0·161***
R850-5500·483R850-550 = 0·259 + 0·000125 × TChl0·171***
R850-6800·488R850-680 = 0·239 + 0·000132 × TChl0·179***
R850/4800·292R850/480 = 31·83 + 0·0140 × TChl0·050**
R850/6800·462R850/550 = 13·46 + 0·0139 × TChl0·107***
R680/850−0·498R680/850 = 0·103 + 0·00000651 × TChl0·122***
δλRE0·335RλRE = 698·8 + 0·00139 × TChl0·069**
δRE0·488RRE = 0·00562 + 0·00000365 × TChl0·147***
PRI0·273PRI = − 0·194 + 0·00000596 × TChl0·059**
NDVI0·498NDVI = 0·813 + 0·000113 × TChl0·125***
CHB0·408CHB = 0·198 + 0·000162 × TChl0·151***
Figure 2.

Relationships between total chlorophyll concentration of moss and the reflectance at 850 nm (a), amplitude of the red edge, δλRE (b), and the NDVI (c). Lines of best fit are exponential curves: R850 = 0·358 + [1 − exp(−0·0099 × TChl)], R2 = 0·352; δλRE = 0·0084 + [1 − exp(−0·0079 × TChl)], R2 = 0·308.

The surface reflectance characteristic of moss that correlated most strongly with the concentration of extracted anthocyanin pigments measured at 526 was δRE (Rho = 0·426). The PRI, a tool developed to remotely assess plant stress (Gamon et al. 1997) was negatively correlated with both the total pool size of the xanthophyll-cycle pigments (Rho = −0·402) and also the proportion as zeaxanthin and antheraxanthin (Fig. 3a; Rho = −0·535). The PRI was positively correlated with the chlorophyll fluorescence parameter, Fv/Fm (Fig. 3b; Rho = 0·626), indicating a declining efficiency of photosystem II.

Figure 3.

Relationships between the photochemical reflectance index (PRI) and the proportion of the xanthophyll-cycle pigments in their active  form  (%A + Z/V + A + Z)   (a)  and   the   chlorophyll   fluorescence  parameter Fv/Fm   (b).  Regression  equations   are:   PRI = −0·0976 + (−0·00209 × %AZ/VAZ), R2 = 0·298; and PRI = −1·285 + (1·604 × Fv/Fm), R2 = 0·427.

To assess the influence of water content on reflectance parameters we rehydrated dry samples of G. antarctici in the laboratory. Re-hydration resulted in a significant increase in δRE accompanied by a shift in δλRE to shorter wavelengths (Table 7). Re-hydration also resulted in a significant increase in the PRI and a decrease in the CHB (Table 7).

Table 7.  Reflectance of the moss Grimmia antarctici when wet and dry. Samples are from ridges (n = 4). Tests are paired t-tests for samples that were dry and then rehydrated for 5 min
  1. NS signifies a non-significant difference at the P=0·10 level.

R320 0·0093 ± 0·0014 0·0104 ± 0·0013NS 
R526  0·026 ± 0·004 0·0022 ± 0·004 −2·620·079
R550  0·031 ± 0·010  0·037 ± 0·004NS 
R850  0·302 ± 0·020  0·294 ± 0·021NS 
δRE 0·0052 ± 0·0001  0·109 ± 0·005−21·120·0002
δλRE  713·25 ± 0·75  699·9 ± 0·41 15·630·006
PRI−0·146 ± 0·042−0·118 ± 0·046 −4·870·016
NDVI  0·881 ± 0·020  0·890 ± 0·018NS 
CHB  0·287 ± 0·021  0·260 ± 0·031  2·920·062



It was found that moss species, moss at different sites, and moss from different microtopographic positions varied significantly in their surface reflectance properties, but different reflectance parameters were sensitive to different ecological or physiological factors (Table 8). The parameters most sensitive to species and site factors were reflectance in the visible region (R526 and R550), and the features of the red-edge (δRE and δλRE). Among moss species, B. pseudotriquetrum, had higher values of reflectance indices than other species (Fig. 1, Table 3). This was not simply associated with differences in chlorophyll concentrations, which were higher than those of C. purpureus but similar to that measured in G. antarctici. Bryum pseudotriquetrum is morphologically distinct from the other two species, having much larger gametophytes with larger leaves that are densely packed into the turf (S.A. Robinson, unpublished results). Gametophyte size is similar in the other two species, but gametophytes of C. purpureus are also densely packed, whereas those of G. antarctici plants are more loosely arranged within the turves. Reflectance parameters are likely to be influenced by gametophyte morphology and packing independently of chlorophyll concentrations. Reflectance from canopies is sensitive to the leaf area index (LAI; Booches et al. 1990; Peñuelas et al. 1994; Serrano, Gamon & Peñuelas 2000) and to other features of canopy geometry, e.g. leaf thickness, surface quality and orientation (Rock et al. 1988). Thus, differences observed in reflectance characteristics among moss species could be more strongly influenced by differences in plant morphology (e.g. gametophyte morphology, density and thickness of photosynthetic tissue) than by physiological factors. It should be noted that B. pseudotriquetrum is the species that shows the most ‘angiosperm leaf like’ reflectance. Unfortunately, as moss surface reflectance has not been well studied we are unable to say if B. pseudotriquetrum or the other species are most typical of moss plants.

Table 8.  Summary of effects of reflectance parameters and pigments over species, sites and microtopography.
ParameterSpeciesSiteMicrotopographyWater content
  1. X indicates a main effect of the factor at P < 0·10, XX at P < 0·05 and XXX at P < 0·001. N/A signifies not applicable because measurements were not made.

 Anthocyanins (A526 diff g−1 FW)   N/A
 Anthocyanins/TChl. (A526 diff mol−1)XX  N/A
 UV absorbing pigments Mean A(320−280nm) (g−1 DW)XXXXXN/AN/A
 Total chlorophyll (nmol g−1 FW)X XXXN/A
 Chl a : b ratio   N/A
 VAZ/TChl (mmol mol−1)XXXXXXN/A
 Tcar./Tchl. (mmol mol−1)XXXXXN/A

In previous work we have found that these three moss species desiccate at different rates and can maintain photosynthetic function at different levels of desiccation (Robinson et al. 2000). Given the sensitivity of the red-edge features to water content (Table 7), differences among reflectance of moss species and sites could also be due to variation in water contents.

Pigment composition also varied between species with B. pseudotriquetrum, having higher levels of specific UV-screening pigments whereas C. purpureus and G. antarctici showed higher levels of pigments generally associated with photoprotection from excess visible light. In the latter species the ratio of xanthophyll cycle pigments/chlorophyll was high and characteristic of plants exposed to much higher levels of solar radiation. Given that samples for pigment analysis were not frozen until several hours after collection, the sustained de-epoxidation status of the xanthophyll cycle pool is also indicative of relatively high photochemical stress in these plants. Differences in relative pigment concentrations between the three species also suggests that they may be differentially vulnerable to increasing UV-B exposure as a result of ozone depletion.


Over different microtopographic positions G. antarctici samples on ridges had lower PRI than samples from valleys, but no differences were observed in any other reflectance parameters, despite large differences in chlorophyll concentrations between moss on ridges and valleys. In other studies higher chlorophyll concentrations and proportion of green tissue are associated with shifts in δλRE to longer wavelengths (Curran, Dungan & Gholz 1992; Vogelmann et al. 1993; Serrano et al. 2000). The lack of difference in the features of the red-edge, or other reflectance indices despite large differences in chlorophyll concentrations over ridges and valleys suggest that δRE and δλRE are largely independent of chlorophyll concentrations over the range found in this study. In both ridges and valleys moss turves form closed microcanopies that probably have high ‘leaf’ area indices (LAI). In higher plants, canopies with high LAI are often not distinguishable using reflectance parameters (Booches et al. 1990; Peñuelas et al. 1994; Serrano et al. 2000). Thus, the dense packing of gametophytes in moss may contribute to the insensitivity of reflectance parameters to differences in microtopography.

Lower PRI on ridges than valleys was associated with higher concentrations of xanthophyll-cycle pigments and greater de-epoxidation of xanthophyll cycle pigments; characteristics of plants experiencing conditions that limit photosynthetic carbon gain and result in photo-inhibition (Osmond 1994; Björkman & Demmig-Adams 1995; Lovelock, Osmond & Seppelt 1995a; Lovelock et al. 1995b). Moss on ridges is exposed to much higher levels of solar radiation than moss which is shaded in microtopographic valleys. Moss on ridges also experience lower and more variable water availability. Higher light levels in combination with lower water contents are likely to lead to photo-inhibition, de-epoxidation of xanthophyll cycle pigments and reduced PRI.


The reflectance indices that were sensitive to site but not species were EGFN, EGFR and CHB. In our assessment of reflectance parameters over sites we expected that sites with lower water availability (Robinson Ridge and Science) would have differences in reflectance parameters and pigment concentrations that were indicative of greater exposure to environmental stress. Our prediction was largely supported with EGFN and EGFR being higher at the drier Science and Robinson Ridge sites, than at the Red Shed site (Table 6). Differences in EGFN and EGFR are products of the differences in the red edge over sites, where the amplitude of the red edge is steeper and at longer wavelengths at the drier sites. The CHB was greater at the cooler western Robinson Ridge site than at Science and Red Shed, which corresponded to the lower average temperatures recorded at Robinson Ridge (Melick et al. 1997).

Total chlorophyll concentrations were greater at Robinson Ridge than at the other sites. Because we did not observe differences in δRE andδλRE due to the microtopographic position of moss, where chlorophyll concentrations were vastly different, we conclude that changes in δRE andδλRE over sites are due to other factors rather than differences in chlorophyll concentrations. Other important factors could be differences in moss species composition over the sites, differences in water content, and/or gametophyte morphology. It is also possible that the higher reflectance at the Science site is due to the increased presence of fine dust particles on the moss surface as a result of local human activity.

Relationships among reflectance, plant pigments, water content and photosynthesis

Despite sensitivity of surface reflectance characteristics to species, site and microtopography, surface reflectance indices were not strongly correlated with pigment concentrations, with the exception of the correlation between PRI and xanthophyll-cycle pigments. The PRI also correlated with the chlorophyll fluorescence parameter Fv/Fm, a measure of photosystem II function, as has also been shown by Gamon et al. (1997) and others (e.g. Méthy 2000). The NDVI, an index commonly used in broad band remote sensing, was shown to be insensitive to the range of factors assessed in this study (Table 8) as has also been observed in other studies (e.g. Buschmann & Nagel 1993). Correlations between reflectance parameters and pigments on a mass basis are expected to be poorer than those on an area basis; however, they have still proved to be useful (see Blackburn 1998).

The water content of moss appears to be a key parameter influencing δRE, δλRE and PRI. The red edge has been shown to shift to longer wavelengths when bean leaves are infiltrated by water (Knipling 1970). More commonly observed is a ‘blue shift’ in leaves and plants that are subjected to water stress (Horler et al. 1983; Booches et al. 1990; Carter 1991, 1993; Peñuelas & Inoue 1999), although in some cases measurements are complicated by a reduction in chlorophyll in the water-stressed tissues (Maracci et al. 1991). Desiccation tolerant mosses show much greater variation in water content than angiosperm leaves (i.e. these species tolerate a 12- to 28-fold difference in water content in live tissues, Robinson et al. 2000). Thus, a shift toward the red as water content increases in moss may reflect processes that are particular to the physiology of desiccation-resistant plants. These processes may be more similar to those involved in greening of angiosperm leaves (e.g. Gates et al. 1965; Curran et al. 1991; Vogelmann et al. 1993) than those as a result of acute environmental stress.

The sensitivity of the surface reflectance indices to species and site and some physiological parameters suggests that mapping of moss communities might be possible using hyperspectral sensors. From these results it seems likely that a high spectral and spatial resolution sensor such as the Compact Airborne Spectrographic Imager (CASI), programmed to detect narrow wavebands in the visible and near infrared, could potentially differentiate between species and detect changes in physiological activity at small spatial scales (1 m2 or less, depending on the deployment height of the sensor). These methods have already proved to be useful in the mapping of aquatic and littoral communities, in which the use of visible wavebands was essential (Zacharias, Niemann & Borstad 1992; Bajjouk, Guillaumont & Poplus 1996; Fyfe & Dekker 2001) and in assessment of photosynthetic activity in terrestrial plant communities (e.g. Zarco-Tejada et al. 2000). Hyperspectral remote sensing could provide an effective and non-destructive method of mapping the moss communities that occur in remote Antarctic and other high-latitude locations, and to monitor impacts of climate change on the spatial extent and composition of these communities that are highly sensitive to climatic trends (Doran et al. 2002).


In Antarctica, moss surface reflectance characteristics vary among species and are significantly influenced by the environmental conditions under which plants grow. Differences in reflectance characteristics across sites and species were not clearly linked to changes in pigment concentrations but may be more strongly linked to differences in water content and morphology of gametophytes and turves. The PRI was the surface reflectance index most sensitive to variation in pigment compositions and photosynthetic function. Hyperspectral imaging of plant communities may be a useful tool in assessing the impacts of global climate change in Antarctica. Moss may also be a useful system to test the applicability of new parameters for remotely sensing the physiological state of vegetation.


This research was supported by grant funding and logistic support from the Australian Antarctic Division. We are grateful for the generous loan of the GBC diffuse reflectance sphere by John Rock, Chemical Engineering, Granville TAFE College. Jane Wasley and other members of the 49–51st ANARE provided field assistance in Antarctica and Johanna Turnbull provided technical assistance with the pigment analysis. We thank Sue Fyfe for her comments on the manuscript.

Received 11 March 2001;received inrevised form 20 May 2002;accepted for publication 20 May 2002