- Top of page
- Materials and Methods
When green plants experience environmental stress (e.g. non-optimal availability of water, light or nutrients), photosynthetic carbon dioxide (CO2) assimilation generally decreases. Even under optimal conditions, plants in all but deeply shaded environments are routinely exposed to light intensities that exceed their capacities for photosynthetic light utilization (Kok, 1956). Environmental stress generally exacerbates excess light stress, as the level of excess light is a function of light intensity and photosynthetic light use (e.g. Logan et al., 1998a,b; for a review, see Björkman, 1981). In cases in which excess light is incident on a leaf, a range of photoprotective mechanisms work to minimize light-mediated cellular damage (for a review, see Melis, 1999).
Thermal energy dissipation, a ubiquitous photoprotective mechanism, depends on the conversion of constituents of the xanthophyll cycle from violaxanthin (V) to antheraxanthin (A) and zeaxanthin (Z) via successive, enzyme-catalyzed de-epoxidations (Yamamoto, 1979). Thermal energy dissipation converts absorbed excitation energy to heat, which can then be lost to the environment across the leaf lamina. Levels of thermal energy dissipation generally increase during exposure to environmental stress (for a review, see Demmig-Adams & Adams, 1996, 2006), and can be regulated on short time scales (i.e. tens of seconds to minutes) via alterations of the xanthophyll cycle de-epoxidation state ([Z + A]/[V + A + Z]) (e.g. Hartel et al., 1999; Peguero-Pina et al., 2013). Chlorophyll fluorescence emission can be used to measure ‘non-photochemical quenching’ (NPQ), which is a common method for the quantification of the level of thermal energy dissipation (Schreiber et al., 1986; Bilger & Björkman, 1990; Müller et al., 2001). Although strong evidence supports Z as the primary pigment responsible for thermal energy dissipation in most plant taxa (Demmig-Adams, 1990; Demmig-Adams & Adams, 1992, 2006), Z-independent forms of thermal energy dissipation have also been described (Bungard et al., 1999; Ruban et al., 2007).
De-epoxidation of the xanthophyll cycle results in a decrease in leaf reflectance between c. 510 and 550 nm centered at 531 nm (Gamon et al., 1990, 1992; Peñuelas et al., 1995; Gamon & Surfus, 1999). This decrease can be detected using remote sensing devices capable of the passive measurement of narrowband (c. 1–10 nm bandwidth) spectral reflectance. Typically, the absorption at 531 nm is used in conjunction with a reference band at 570 nm that is insensitive to changes in xanthophyll cycle de-epoxidation state to calculate a normalized index, known as the photochemical reflectance index (PRI) (Eqn 1) (Gamon et al., 1992):
- (Eqn 1)
Strong correlations have been reported between PRI and metrics associated with the xanthophyll cycle (de-epoxidation state, NPQ) at the leaf and canopy scales (e.g. Gamon et al., 1990; Gamon & Surfus, 1999; Richardson & Berlyn, 2002; Filella et al., 2004b; Rahimzadeh-Bajgiran et al., 2012). Thus, owing to their spectral reflectance signal, changes in xanthophyll cycle composition can be tracked via passive remote sensing platforms from the leaf to ecosystem scale (see reviews by Garbulsky et al., 2011; Peñuelas et al., 2011).
Although much progress has been made on the passive, remote detection of plant responses to environmental stress over the last several decades (Barton, 2012), many challenges associated with the use of the PRI persist. These challenges include accounting for canopy structural differences and changes (e.g. in leaf area, leaf orientation and leaf angle distribution; Barton & North, 2001; Garrity et al., 2010), background effects (e.g. woody or dead canopy material, soil, shadows; Barton & North, 2001) and differences in leaf chlorophyll or carotenoid pools (Filella et al., 1996; Sims & Gamon, 2002; Stylinski et al., 2002; Garrity et al., 2011) – although advances at a variety of spatial scales have been reported. These recent advances have allowed for the inference of photosynthetic performance of forest canopies by accounting for viewing and illumination geometry through multi-angle measurements that relate the rate of change in PRI to canopy shadow fraction (e.g. Hall et al., 2008, 2011, 2012; Hilker et al., 2008, 2009, 2010, 2011, 2012). By building on these recent advances, some of the challenges associated with the passive remote sensing of PRI may be further overcome with the use of active remote sensing systems, such as terrestrial laser scanning (TLS, see the 'Discussion' section).
Traditionally, TLS technology has been used to map and model plant structural properties (e.g. Lovell et al., 2003; Hopkinson et al., 2004; Watt & Donoghue, 2005; Clawges et al., 2007) by the quantification of the x, y, z locations of plant canopy components at very fine (sub-centimeter) resolution. Locations are calculated using the angle-specific round-trip time-of-flight (t) of a laser pulse between the sensor and the target (distance = (ct)/2, where c is the speed of light). In addition, the TLS measures the return intensity of the returning laser pulse. Because the 532-nm laser is centered within the spectral region known to be affected by de-epoxidation of the xanthophyll cycle, we hypothesize that this wavelength could tease apart the three-dimensional (3-D) variability in the levels of thermal energy dissipation at the leaf and canopy scale. Following the theoretical work of Morsdorf et al. (2009) and Woodhouse et al. (2011) – who demonstrated the potential for measuring detailed forest structural and physiological status using light detection and ranging (LiDAR) through modeling exercises – we present the first attempt to characterize changes in the xanthophyll cycle using green scanning LiDAR.
Although research to determine vegetation structure with terrestrial LiDAR systems is rapidly expanding (e.g. Keightley & Bawden, 2010; Moorthy et al., 2011; Sanz-Cortiella et al., 2011; Vierling et al., 2012; Fernández-Sarría et al., 2013), research examining the potential of LiDAR systems to determine plant biochemical properties is just emerging (Eitel et al., 2010, 2011; Gaulton et al., 2013). Recent research by Eitel et al. (2010, 2011) suggests that laser scanning technology could be a useful tool to accurately map biochemical properties at both the leaf and canopy scale. They showed that the laser return intensity of a green (532-nm) scanning laser is significantly correlated with foliar chlorophyll content (Eitel et al., 2010) and total foliar nitrogen (Eitel et al., 2011). In addition, work by Gaulton et al. (2013) has found that dual-wavelength laser scanning using near-infrared (1063-nm) and middle-infrared (1545-nm) wavelength lasers can be employed to estimate leaf moisture content. However, there has yet to be any published research demonstrating the use of a TLS system to measure dynamic changes in leaf physiological state.
The objective of this work was to test our hypothesis that variations in reflected green laser return intensity (GLRI) at 532 nm, measured by TLS, can track temporal changes in thermal energy dissipation at the leaf scale. We assessed the relative abilities of GLRI and PRI to predict NPQ and the xanthophyll cycle de-epoxidation state. This study was conducted at the leaf scale under controlled laboratory conditions to obtain a fundamental understanding of the suitability of GLRI for the assessment of leaf photoprotective mechanisms. Following a report on our findings at the leaf scale, we discuss prospects for scaling the TLS intensity data to a 3-D canopy space.
- Top of page
- Materials and Methods
The near-equivalent performances of GLRI and PRI in the prediction of xanthophyll cycle-dependent light energy dissipation (Figs 5-7) suggest that laser-based intensity measurements (such as GLRI) can be used to assess dynamic changes in xanthophyll cycle de-epoxidation at the leaf scale. This finding is a fundamental advance that supports previous foundational and theoretical work (Morsdorf et al., 2009; Woodhouse et al., 2011; Wallace et al., 2012), suggesting that laser measurements might be useful in ascertaining foliar physiological information in three dimensions. In addition, several patterns in the tracking of leaf physiological properties using active and passive remote sensing emerged from this work that should not be overlooked in future scaling exercises.
First, we noticed that, as chlorophyll content increased in leaves (a result of differing nitrogen treatments), the relationships between GLRI/PRI and NPQ/((Z + A)/(V + A + Z)) improved. This could potentially be explained by the strong correlation between bulk xanthophyll cycles (V + A + Z) and chlorophyll (a + b) content on a leaf area basis (μmol m−2) (r = 0.76, across all species in this study). Because the leaves used in this study were grown in low light conditions and mostly absent from situations of high light stress, leaves with low chlorophyll levels may not have needed to develop large xanthophyll cycle pigment pools, which could be superfluous in situations in which a high capacity for photoprotection is not routinely required. The poor PRI/GLRI response in leaves with low chlorophyll content (and concurrently low xanthophyll cycle contents) may be the result of the low optical detectability of a small physiological response. It is worth noting that a relatively large difference in xanthophyll cycle de-epoxidation or NPQ corresponds to a narrow range of differences in PRI or GLRI, which means that a small variation or error in the latter could translate into a larger difference in physiological measures. The above finding is consistent with the observations of other studies (Gamon et al., 2001; Filella et al., 2004a,b; Nakaji et al., 2006; Garrity et al., 2011; Rahimzadeh-Bajgiran et al., 2012), which demonstrate that the relationship between PRI and the xanthophyll pigment interconversion is weaker in leaves with low chlorophyll content. On longer time scales (days to weeks), PRI responds to variables other than xanthophylls (i.e. chlorophyll/carotenoid ratio) and, as a result, it is necessary to correct PRI (or GLRI) measurements by incorporating the effect of other photosynthetic pigments (Sims & Gamon, 2002; Filella et al., 2009; Garrity et al., 2011; Rahimzadeh-Bajgiran et al., 2012).
Second, in leaves in which concentrations of Z and A, and levels of NPQ, were higher (generally sunflower, saplings, wheat – in that order), GLRI and PRI performed relatively better as predictor variables. This could simply be explained by the wider range of xanthophyll de-epoxidation states triggered by species-specific responses (see Figs 6, 7), which, in turn, would induce a wider range in spectral variation at 532 nm. Another explanation could be that our strongest results come from the thinnest, flattest leaves under study; although not the aim of this research, further work should be performed to test how GLRI could be affected by leaf structure (rigidity, leaf lamina) and cellular structure (leaf mesophyll and epidermis – see Ollinger, 2011 for further discussion).
A ‘finer frontier’ for remote estimation of photoprotection
Great strides have been made over the past two decades to better quantify net ecosystem exchange (NEE) of CO2 (e.g. Baldocchi, 2008). However, the complexity of observing and modeling NEE over broad areas remains a challenge because carbon exchange is highly variable in time and space (Le Quéré et al., 2009). The documentation, understanding and forecasting of the cycling of carbon in response to global environmental change therefore rely heavily and increasingly on remote sensing systems (Running et al., 2004). Because the modeling and prediction of carbon sources and sinks at leaf, canopy and ecosystem scales require a nuanced understanding of vegetation structure and physiology (Schurr et al., 2006), remote sensing approaches must be responsive to physiological and structural vegetation changes that are relevant to carbon uptake by photosynthesis (Ustin & Gamon, 2010). A recent letter (Peñuelas et al., 2011) and Tansley review (Ustin & Gamon, 2010) published in this journal identified key challenges associated with passive remote estimates (namely, the PRI) of plant stress and function. Some of these challenges include accounting for canopy structural differences and changes, viewing and illumination geometry, background effects, changes in illumination conditions, and poor spectral and/or spatial resolution. An active, laser-based system could work to minimize these challenges.
First, a major challenge of passive optical remote sensing has been the inability to understand that plant spectral reflectance resulting from changes in leaf pigment activity varies among functional types (Gamon et al., 1997), and is complicated by vegetation structure (Barton & North, 2001; Middleton et al., 2009). Because leaf spectral response is often dictated by both canopy biochemical (chlorophyll, nitrogen, xanthophylls) and biophysical (leaf angle distribution and orientation) status, difficulties in interpreting the data from remote sensing platforms can arise. As a potential solution, the high-resolution imaging capability of vegetation 3-D structure with TLS, which can be used to determine leaf angle distribution and orientation, coupled with dynamic 3-D changes in biochemical conditions, could provide new insights into spatially explicit patterns of physiological function on the leaf scale (Figs 3, 4) and the canopy scale (as discussed in the following section).
Second, Hall et al. (2008) first demonstrated that differences in canopy PRI resulting from differing viewing and illumination geometries can be used to detect xanthophyll-induced changes at 531 nm. This was accomplished using remote sensing instruments and techniques that allow canopy spectra to be recorded at multiple angles (i.e. multi-angular spectroradiometer platforms capable of near-ground measurements; Hilker et al., 2009, 2010). A more mechanistic understanding of the relationship between canopy function and structure issues may be accomplished by coupling multi-angle imaging spectroscopy with LiDAR, thereby isolating subtle variations in vegetation reflectance associated with changes in xanthophyll cycle de-epoxidation. In addition to fusing passive spectral and active LiDAR data at the canopy scale to understand canopy function, there is wide interest in exploring ‘single-system’ techniques (such as TLS), which have the potential to improve upon the 3-D spatial understanding of leaf and canopy biochemistry (Omasa et al., 2007; Eitel et al., 2010, 2011; Gaulton et al., 2013). Green scanning TLS could eventually aid in research aimed at unraveling the ‘canopy conundrum’ (see Grace, 2007; Nichol et al., 2012), whereby the parameterization of vegetation structure and function at the leaf scale could elucidate the large-scale photosynthetic processes controlled by canopy and ecosystem structure. For example, a TLS system could provide a vertical, spatially referenced, high-resolution profile of dynamic photosynthetic changes through a tall tree canopy.
Third, varying background effects (e.g. soil color, non-photosynthetic material) may considerably alter passive estimates of photosynthesis (e.g. Sims et al., 2006). The fine spatial resolution of TLS exemplified in Fig. 4 can help account for these effects, because the exact 3-D location of a particular reflectance measurement is known. Background laser returns coming from soil or non-photosynthetic material can also be easily separated from vegetation laser returns by the use of easily applied thresholds (Eitel et al., 2011).
Fourth, the advantage of lasers over traditional passive sensors is that the laser return intensity is not affected by ambient light conditions, and can be measured in the dark to acquire a baseline measurement of foliar reflectance at 532 nm (i.e. in the absence of photosynthesis). Further, the high sampling rate (50 000 points s−1 for this particular terrestrial laser scanner), combined with the small field of view of lasers (< 4 mm), allows pure green vegetation returns to be isolated from non-photosynthetic tissue (Eitel et al., 2010, 2011). This permits the measurement of the 532-nm laser return intensity signal from individual leaves and portions of leaves growing within a radius of up to 150 m from the laser source (in the case of the instrument used in this study).
Achieving three dimensions: considerations and limitations for the use of GLRI at the canopy scale
When implementing GLRI to study plant physiological status at the canopy scale, the interpretation of laser return intensity values might be confounded by returns from non-photosynthetic tissue, variations in leaf angle and the distance between the laser and the surveyed leaves (Eitel et al., 2010). To isolate intensity readings from non-photosynthetic tissue, a simple threshold can be used (Eitel et al., 2010, 2011). The distance readings provided by TLS can allow the inverse distance square law to be accounted for, whereby the intensity of light returned to the laser decreases as a function of the square of increasing distance from the surveyed object. A greater challenge for the implementation of TLS measurements to infer photosynthetic performance at the canopy scale involves accounting for leaf angle effects on laser return intensity. To minimize the confounding effects of leaf angle in this study and establish that GRLI can be used to derive leaf photoprotective status, leaf angle was kept constant. Hence, our results effectively represent two-dimensional measurements. However, in order for GLRI to provide novel 3-D insights into photosynthetic performance throughout plant canopies, techniques and methods are required that allow variations in leaf angle to be accounted for. Although recent work has successfully calculated the leaf angle using TLS data (e.g. Eitel et al., 2010; Balduzzi et al., 2011; Zheng & Moskal, 2012), this work has not yet produced algorithms sufficient to calculate the leaf angle automatically. Because automatic leaf angle calculation is essential for scaling our work to a typical vegetation canopy containing > 102 leaves, we are developing a robust, nearest-neighbor-based algorithm to automatically calculate the leaf angle of every laser return recorded by TLS (Fig. 8; J. Jiang et al., unpublished). Based on the leaf angle data provided by such an algorithm, laser return intensity values could be corrected for variations in leaf angle to accurately map changes in NPQ and xanthophyll pigment pools in three dimensions.
Figure 8. Leaf angle normals of (a) oak, (b) wheat leaves and (c) an entire wheat canopy. Lines emanating from each point on the leaves depict the leaf angle normals for each individual laser hit on the leaf surface, as calculated using an automated nearest-neighbor surfacing algorithm (J. Jiang et al., unpublished). Because the leaf angle can exert a strong control on the green laser return intensity (GLRI), this leaf angle information offers new prospects for scaling measurements made in the laboratory to actual field conditions in three-dimensions.
Download figure to PowerPoint
The use of TLS systems employing multiple wavelengths (e.g. Strahler et al., 2008; Jupp et al., 2009; Morsdorf et al., 2009; Chen et al., 2010; Woodhouse et al., 2011; Hakala et al., 2012; Gaulton et al., 2013) could also help to reduce the effect of leaf angle and distance on laser return intensity values. By calculating ratios of laser returns, laser return intensity readings could be normalized for changes in leaf angle and distance, similar to widely employed spectral vegetation indices. Gamon et al. (1992) originally found that a wide range of reference bands accompanying the 531-nm ‘xanthophyll wavelength’ could be used to minimize the effects of overlapping spectral features on the ‘xanthophyll signal’ caused by sun and leaf angle variation, although 570 nm has been the most widely used wavelength in PRI calculations (see Garbulsky et al., 2011 for a review). Future developments in laser technology could aid in exploring the use of a reference band in active remote sensing of plant photoprotective mechanisms.
The power of active remote sensing data is that they eliminate the confounding effects of shadow and non-photosynthetically active elements that confound passive remote sensing metrics; however, another limitation inherent in any remote sensing approach is that a fixed view angle will not ‘see’ objects between the scanner and the outer leaves (as described in the work of Hall et al., 2008). By scanning the plant canopy from different viewing positions (e.g. Clawges et al., 2007), more photosynthetically active materials could be ‘seen’, providing a more complete picture of canopy photosynthetic performance. Notably, because photoprotective mechanisms respond to foliar-level illumination conditions (i.e. shaded leaves experience lower ‘light stress’ than sunlit leaves), the interpretation of canopy-level GLRI signals would be assisted by illumination data specific to the time of measurement. Because most TLS systems are outfitted with an onboard red–green–blue (RGB) digital camera, the digital image could be used to map instantaneous canopy shadow conditions to better interpret the GLRI variation on a leaf-by-leaf basis. Combination of the rapid scan time and RGB digital imaging afforded by TLS systems could aid in the development of a multi-angle data acquisition approach that accounts for shadow fraction, whereby TLS scans and digital images are acquired for the same canopy from multiple angles over a short time frame. Finally, as TLS data have recently served as the basis for the development and evaluation of advanced 3-D geometric-optical ray tracing models of canopy illumination (e.g. Bittner et al., 2012), these models could be implemented to calculate foliar illumination conditions, offering further interpretation power for understanding leaf-level GLRI variation as it relates to foliar photoprotective physiology.
Because current methods of TLS data acquisition require an instrument operator to be present during scan acquisition, new opportunities to monitor variation in canopy 3-D photosynthetic function over time would be afforded by the development of automated LiDAR scanning technologies. Indeed, new techniques enabling reliable, repeatable LiDAR scans under a wide range of environmental conditions could greatly improve the operational capacity and interpretation of GLRI canopy physiology data. Recently, Eitel et al. (2013) developed a lightweight, low-cost, autonomously operating near-infrared laser scanner to quantify and monitor ecosystem structural dynamics; this concept of automated scanning is readily applicable to a green laser in order to acquire GLRI.
By accounting for leaf angle, viewing angle and the need for continuous data, we feel that the GLRI results shown here hold promise for future inferences of canopy photosynthetic performance in three dimensions. In addition to the impending advances afforded by the utilization of TLS data in three dimensions, our work demonstrates that spatial variation in foliar photoprotection using GLRI can be further explored in two dimensions at extremely high spatial resolution (Figs 3, 4). Seminal work towards the mapping of the efficiency of photosynthetic light utilization has already been addressed using a laser-induced fluorescence transient (LIFT) method for terrestrial vegetation (e.g. Ananyev et al., 2005; Kolber et al., 2005). In addition, 2-D measurements coupled with 3-D LiDAR measurements (Omasa et al., 2007; Hilker et al., 2008; Konishi et al., 2009) are an important step towards the better quantification of canopy-level photosynthetic parameters. In sum, although work remains to implement the findings of this paper for the study of 3-D canopy photosynthetic performance in situ, our results underscore the applicability of LiDAR technology as a viable and critical tool for the investigation of previously unobservable spatiotemporal patterns and relationships between plant structure and function.