Photochemical Reflectance Index (PRI) captures the ecohydrological sensitivity of a semi-arid mixed conifer forest

The Photochemical Reflectance Index (PRI) corresponds to the de-epoxidation state of the xanthophyll cycle and is one of the few pigment-based vegetation indices sensitive to rapid plant physiological responses.  As such, new remotely-sensed PRI products present opportunities to study diurnal and seasonal processes in evergreen conifer forests, where complex vegetation dynamics are not well reflected by the small annual changes in chlorophyll content or leaf structure.  Because PRI is tied explicitly to short and long term changes in xanthophyll pigments which are responsible for regulatig stress, this study characterized PRI in a semi-arid, sub-alpine mixed conifer forest, in order to assess its potential as a proxy for water stress by extension of its association with photoprotection.  To determine the sensitivity of PRI to seasonal changes in ecohydrological variability and gross primary productivity, canopy spectral measurements were combined with eddy covariance flux and sap flow methods.  Seasonally, there was a significant relationship between PRI and sap flow velocity (R2=0.56), and multiple linear regression analysis demonstrated the PRI response to dynamic water and energy limitations in this system.  Although PRI was an effective indicator of stomatal response to ecohydrological constraints on a seasonal time scale, top-of-canopy leaf-level gas exchange, chlorophyll fluorescence, and hyperspectral reflectance measurements suggest that diurnal PRI saturates under conditions of severe light stress.  This research indicates that remotely-sensed PRI has potential to fill spatial and temporal gaps in the ability to distinguish how water availability influences carbon dynamics of forested ecosystems.
Methods
Study Site:  The study location was a sub-alpine mixed conifer forest in the Coronado National Forest on Mt. Bigelow, northeast of Tucson, Arizona. The site is at 2573 m elevation in an area of significant topographical complexity.  The climate is semi-arid with a mean annual temperature of 9.4 °C andmean annual precipitation of 750 mm. Of this, ~50% falls during the North American Monsoon in late summer (Adams et al., 1997).  The site is dominated by mature second-growth Douglas fir (Pseudotsuga menziesii), ponderosa pine (Pinus ponderosa), and southwestern white pine (Pinus strobiformis), withlittle understory vegetation. The forest exhibits a complex and bimodal pattern of primary production, with an initial spring peak following snow melt, a dry pre-monsoon mid-season depression (May-June), and a secondary productivity peak during the wet monsoon (July-Sept), remaining active through fall. 

Sap flow:  We measured sap flow on the north and south sides of three P. strobiformisand two P. ponderosaindividuals using the thermal dissipation probe method (Granier 1985; Granier 1987).  Data were logged at 30 min resolution using an upper heated probe and lower reference probe (TDP-30, Dynamax Inc., Houston, TX, USA) implanted in the sapwood of the tree approximately 40 mm apart.  Sap flow velocity (cm hr-1) was calculated according to:    , where  and dT is the difference in temperature (°C) between the two needles, and dTM is the maximum temperature difference between midnight and 7:00 am.  

Canopy Spectral Reflectance:  On July 3, 2018, we installed an autonomous Spectral Reflectance Sensor (METER Group, Inc., Pullman, WA, USA), and began collecting PRI reflectance at 10-min intervals.  For a complete description of the sensor see: http://manuals.decagon.com/Manuals/14597_SRS_Web.pdf and Magney et al., 2016.  The PRI sensors use photodiodes with narrow bandpass filters centered at the 532 nm and 570 nm wavelengths with 10 nm full width half maximum bandwidths. It uses a hemispherical upward-looking sensor, and a field stop downward-looking sensor to measure incoming and upwelling radiation (W m-2sr-1nm-1), respectively.  PRI was calculated as:                

Where  is the spectral reflectance value at center wavelength of 532 nm and  is the spectral reflectance value at center wavelength of 570 nm.  Downward looking sensor interference filters restrict the field of view (FOV) to 36°.  The sensor was at 24 m height, roughly 12 m above the top of the canopy titled off-nadir at an angle of 20°, resulting in a field of view (FOV) of ~50m2. The PRI sensor faced west and therefore measured eastern facing needles.  Within the sensor FOV were full or partial canopies of five trees (three P. ponderosaand two P. strobiformis, no understory vegetation), four of which were equipped with sap flow sensors.

Leaf Level gas exchange, chlorophyll fluorescence, and hyperspectral reflectance:  We collected leaf-level measurements on September 13-14 for one P. ponderosaand one P. strobiformismature tree on attached top of canopy needles (13m height) using a canopy access crane.  We measured four branches on each tree every hour from 9:00 -16:00 MST.  Two sunlit fascicles were measured for simultaneous gas exchange and fluorescence (see below).  Immediately after, the same needles, plus two more fascicles, were measured with a spectroradiometer for PRI (see below).  When measuring under intermittent cloudiness, measurements were aborted if the needles were not exposed to sunlight immediately prior to both gas exchange and spectral measurements. 

Spot gas exchange and simultaneous chlorophyll fluorescence were measured using the Li-6800 Portable Photosynthesis System infrared gas analyzer (LICOR Inc., Lincoln, NE, USA) with a standard 6 cm2 leaf chamber.  For each round of branch measurements air temperature (Tair) and photosyntheticall active radiation (PAR) were characterized, and the internal chamber conditions were set to match the ambient environment.  We performed leaf area analysis on ten samples of each species to derive an average leaf area within the chamber (2.24 cm2 ±0.16 cm2and 2.22 cm2 ±0.22 cm2 for P. ponderosaand P. strobiformis, respectively) assuming each sample clamped the same approximate amount of leaf area.  We used the multiphase flash method to obtain greater accuracy in Fm’ acquisition compared to typical rectangular flash methods (Loriaux et al. 2013).  To obtain dark adapted parameters, rectangular flash measurements were taken at pre-dawn on September 14 for a minimum of eight samples per branch to obtain branch averaged Fo  and Fm. 

We measured leaf-level hyperspectral reflectance with an ASD FieldSpec3 (ASD Inc., Boulder, CO, USA) spectroradiometer with plant probe.  The plant probe has a low intensity light source for non-destructive data collection. Prior to each measurement, a white reflectance reference was taken using a calibrated Spectralon reference standard.  Needles were arranged in a single plane to minimize gaps without overlapping (Rajewicz et al., 2019).  After clamping onto the needles and turning on the light source, 2-3 spectra were taken within a few seconds to prevent jumps in PRI due to an altered light condition (Mottus et al. 2017).