SEARCH

SEARCH BY CITATION

Keywords:

  • Fagus crenata;
  • Vcmax;
  • broadband;
  • vegetation indices;
  • seasonality

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

[1] In this study, we investigated the relationship between meteorological-based broadband simple ratio (SR) and canopy photosynthetic capacity (maximum carboxylation rate normalized to 25°C, Vc,25) in three Fagus crenata stands in the cold-temperate zone of Japan. Broadband SR was calculated from recorded up- and down-looking photosynthetically active radiation (PAR) and global radiation (GR) data. The study reveals that broadband SR seasonal courses closely follow Vc,25 seasonal trajectories, with a statistically significant linear regression relationship between the two. Linear regression models show that R2 ranges from 0.59 (for 550 m site) to 0.91 (for 1500 m site), but eventually drops to 0.37 when all data pool together. Despite variations in R2 for the different sites, the relationship remains statistically significant (P < 0.000). Though spatially limited, broadband SR can serve as an easy but robust indicator of seasonal variations in Vc,25 required for accurate carbon fixation simulations in gas exchange models.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

[2] Canopy photosynthetic capacity indicates the potential of canopy photosynthesis and is a crucial factor in understanding the carbon cycle of a forest ecosystem. Maximum carboxylation rate (Vcmax) normalized to 25°C (Vc,25) is a major canopy photosynthetic capacity parameter, and is a required input in most gas exchange models. Increasing evidence suggests that the parameter Vc,25 is an underlying factor for large seasonal variations in both deciduous forest ecosystems [e.g., Wilson et al., 2000; Xu and Baldocchi, 2003; Wang et al., 2008] and evergreen forest ecosystems [e.g., Tu, 2000; Nogues and Alegre, 2002]. However, few models have included seasonal variations of this crucial parameter into gas exchange models, due to tremendous time and man-power requirements for monitoring over seasonal courses by traditional gas exchange techniques. On the contrary, most modelers grossly ignore seasonal variations in Vc,25, leading to large simulation discrepancies of more than 50% [Wilson et al., 2001]. It is therefore imperative that alternative methods that can speedily estimate Vc,25 seasonal trajectories and are compatible with gas exchange models be developed if improvements in the simulation performance are to be achieved.

[3] Earlier studies suggest that both satellite-based and meteorological data-based vegetation indices may bear close links with seasonal trajectories of canopy physiological activity [Wang et al., 2004]. Calculations of broadband vegetation indices from both incoming and reflected tower-mounted photosynthetically active radiation (PAR) and global radiation (GR) above tree canopies were first proposed by Huemmrich et al. [1999] and subsequently applied in their four BOREAS (Boreal Ecosystem-Atmosphere Study) sites. Wang et al. [2004] followed the proposed method by Huemmrich et al. [1999], and found a close relationship between broadband normalized difference vegetation index (NDVI) and gross primary production (GPP) in a Scots pine forest (Pinus sylvestris L.). In contrast to their remote sensing counterparts, broadband vegetation indices are insensitive to weather and therefore can be used to monitor entire seasonal courses of canopy activity. In this sense, broadband vegetation indices are superior over satellite-based vegetation indices, which often show serious defects from the effects of different weather conditions.

[4] In this paper, we investigated the relationship between meteorological data-based broadband simple ratio (SR) and canopy photosynthetic capacity (Vc,25) in three contiguous natural beech forest (Fagus crenata) stands at different altitudes in the cold-temperate zone of Japan.

2. Data and Methods

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

2.1. Study Area

[5] Beech (Fagus crenata) is a late successional and climax species regarded as the most ecologically important forest tree species of the cool-temperature natural forests in the mountain area of Japan [Ishizuka, 1974]. The study area is located in the Naeba Mountains, Japan (36°51′N, 138°40′E), where beech forests are widely distributed, from the lowest to the highest altitudes, on the northern slope of the mountain. Three towers have long been set in three typical stands representing the lowest, optimum and highest limits of beech distribution at 550 m, 900 m, and 1500 m, respectively. All the sites are dominated by Fagus crenata, with occasional occurrences of other species like Quercus mongolica var. grosseserrata, Magnolia obovata and Acanthopanax sciadophylloides at the 550 m and 900 m sites, and Betula grossa and Betula ermanii at the 1500 m site. Dense Sasa kurilensis understory exists at the 1500 m site. In 2006, leaf area index (LAI) estimates from litter traps at 1.0 m above the ground were respectively 4.15 at the 550 m site, 5.52 at the 900 m site and 3.47 at the 1500 m site. The whole region receives large amounts of precipitation, of about ca. 2000 mm/yr, with long periods of snow cover. Mean annual air temperatures are generally low. For instance, in 1999, recorded mean annual air temperatures were 10.0, 9.3 and 5.6°C at the 550 m, 900 m and 1500 m sites, respectively. At the 550-m and 900-m sites, beech leaves begin to flush in late April to early May, and autumn leaf coloring starts in late October. At the 1500 m site, leaf flushing starts in late May to early June, and autumn leaf coloring begins in early October. On the average, the 550 m site has a growth span of 193 (±8) days while growth span for the 1500 m site is observed to be only 144 (±5) days.

2.2. Radiation Data

[6] Radiation data were measured using two pairs of global radiation and photosynthetically active radiation (PAR) sensors in an up- and down-looking configuration over tree canopies, typically 4 ∼ 5 meters above the canopy. The radiation sensors used for global radiation (both incoming and reflected) were LI-200SL pyranometer (LiCor, Lincoln, NE, USA), and the LI-190SL quantum sensor (LiCor, Lincoln, NE, USA) was used for the PAR. The pyranometers have a wavelength range of 0.3 ∼ 4.8 μm, while PAR sensors are 0.4 ∼ 0.7 μm in wavelength. Half hour time-step data were retrieved by averaging outputs at 30-min intervals. For the conversion of photon flux density (μmol/m2/s) into energy flux density (W/m2), Ross and Sulev [2000] conversion factors were used, with 0.2195 for incoming PAR and 0.2072 for reflected PAR.

2.3. Broadband Reflectance and Simple Ratio Calculations

[7] PAR reflectance, that is, reflected and incoming PAR ratio from the up- and down-looking sensors was calculated as:

  • equation image

Meanwhile optical infrared radiance reflectance, that is, reflected and incoming ratio of irradiance difference between GR and PAR was calculated as:

  • equation image

Following the concept of its narrowband counterpart, we defined the broadband SR as the ratio of infrared to PAR reflectance and quantified it as follows:

  • equation image

Broadband SR is similar to the ground-based normalized difference vegetation index (NDVI) proposed by Huemmrich et al. [1999], but was preferred in this study since it is more sensitive and more linearly responsive to bio-parameters [Chen and Cihlar, 1996].

2.4. Gas Exchange Measurements and Vc,25 Calculations

[8] Gas exchange measurements were mainly based on detached samples under controlled laboratory conditions in portable gas exchange systems (LI-6400, LI-COR, Inc., Lincoln, Nebraska, USA). Sample collection was done at a dense frequency, typically 3 days to 2 weeks, in both the leaf-expansion and leaf-senescence periods in order to capture the rapid changes in photosynthetic capacity. However, in the leaf-constant period (from late June to early September), monthly sampling was adopted. In each sampling, at least 3 to 5 top canopy leaf samples were collected. And in each leaf sample, an immediate determination of photosynthesis to intercellular CO2 concentration (A/Ci) response curve was done at 8 to 10 Ci levels with an ambient CO2 concentration of 0 ∼ 2000 μmol/mol.

[9] Vcmax was then fitted from the A/Ci curve data points at Ci < 250 μmol/mol, the level at which it is generally accepted that A is limited solely by Vcmax in the Farquhar inverse model [Farquhar et al., 1980]. Details have been given by Wang et al. [2008]. In order to limit the effect of seasonal variation in temperature, we applied only Vcmax measurements under a controlled temperature of 25°C in this study. We then also compared Vc,25 standardized from another Vcmax data set measured by an open gas exchange system (Compact Minicuvette System, Walz, Effeltrich, Germany), where chamber temperature was adjusted to the mean monthly air temperature corresponding to the measurement date, before converted to Vc,25 based on the temperature response equation of Harley and Tenhunen [1991]. Both data sets are similar with a maximum difference of <4.6% for all the three sites.

2.5. Regression Model

[10] Broadband SR data for the days on which leaf sampling was done were selected from daily time-step series, which are daily averages of daytime half-hour SR. Finally, regression analysis was done to evaluate the broadband SR and Vc,25 data pairs.

3. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

[11] The broadband SR and Vc,25 seasonal courses for 2006, for all the three sites, are shown in Figure 1. Seasonal trajectories of both broadband SR and Vc,25 are generally similar, with the greatest similarity in seasonal pattern occurring in site 1500 m. Despite some irregular abrupt changes in broadband SR seasonal course, it takes the general form of an inverted parabola with peak values occurring in mid growing seasons, which is similarly the case for Vc,25. Broadband SR varies from 1.9 ∼ 27.4 in the 550 m site, 4.8 ∼ 36.4, in the 900 m site, and 1.3 ∼ 18.5 in the 1500 m site. For the three sites, daily broadband SR amplitudes were minimal in the spring leaf-expansion and the autumn leaf-senescence periods in comparison to the relatively maximal amplitudes in the summer leaf-constant period. Vc,25 peak values for the three sites were similar, with 61.1, 57.5 and 55.2 μmol/m2/s for the 550, 900, and 1500 m sites respectively. Peak occurrence times were very different, which was July 4 in the 550 m site, September 3 in the 900 m site and July 8 in the 1500 site. Peak value occurrence times had direct influence on Vc,25 seasonal patterns.

image

Figure 1. Seasonal courses of broadband SR and Vc,25 for all three sites in 2006.

Download figure to PowerPoint

[12] From the start of the growing season up to full leaf-expansion period, broadband SR pattern closely followed that of Vc,25 in all three sites. However, some site-specific discrepancies were spotted for the other growth periods. For the 550 m site, a large discrepancy occurred in the autumn leaf-senescence period, when Vc,25 continuously dropped till October 1, 2006, after which period a small rebound was noted. There was hardly any distinguishable discrepancy in the pattern of the broadband SR time series. For the 900 m site, two data values in Vc,25 trajectory distorted the similarity between seasonal course with broadband SR. One was an unusually low value on June 21, 2006, and the other was an unusually high value on September 3, 2006. We suspect that the high amplitude may have been induced by sampling errors. Although there was a relatively higher noise in broadband SR time series in the 1500 m site, very close patterns were, nevertheless, noted for both broadband SR and Vc,25.

[13] Figure 2 is an illustration of a scatter diagram of Vc,25 against broadband SR, with linear regression analysis for each site. The statistical results revealed significant linear relations (P < 0.000) between Vc,25 and broadband SR in all three sites, with R2 of 0.59 for the 550 m site, 0.72 for the 900 m site, and 0.91 for the 1500 m site (see Table 1). There was still a statistically significant relationship when all data pooled together, even though R2 dropped to 0.37, which underlines the importance of site effect on the relationship between Vc,25 and broadband SR.

image

Figure 2. Scatter diagram of broadband SR and Vc,25 for all three sites.

Download figure to PowerPoint

Table 1. Vc,25 and Broadband SR Linear Regression Coefficients: Regression Analysis for All Data Pooling Together on Site-Specific Basisa
SiteVc,25 = a + bSR
abR2FP
  • a

    The form of the linear regression model is Vc,25 = a + bSR, where Vc,25 is in μmol/m2/s.

550 m−0.72032.18080.5923.3.000
900 m−0.05431.59510.7231.0.000
1500 m−4.28513.93890.91105.8.000
All14.10001.25800.3724.3.000

4. Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

[14] There were considerable variations in broadband SR and Vc,25 slopes across the three sites (Table 1). A large scatter of the data points was also evident in all the sites (Figure 2). This may partly be due to differences in LAI for different site since SR was low with low LAI in site 1500 m and high with high LAI in site 900 m. In order to isolate the possible site effects on the relationship between broadband SR and Vc,25, we applied ANCOVA analysis, using site as a group factor. The results revealed significant altitude effect on broadband SR-Vc,25 relationship (P = 0.002 < 0.05). When site effects were eliminated, R2 improved to 0.62.

[15] The 550 m site had the lowest R2 among all the sites. The relatively poor performance in the 550 m site may probably be attributed to the down-looking sensors at the site. Although down-looking sensors recorded reflected PAR and GR at 4 ∼ 5 meters above canopy heights, there was large forest gaps in close proximity with the selected canopy in the 550 m site. Furthermore, some shrub species and beech seedlings stood in the gaps. Being cosine receivers, both PAR and GR sensors received reflected signal from the understory canopies beside tree canopies. Vc,25 was only analyzed at tree canopy level, which explains the autumn discrepancies – when physiological processes in the trees and understory species were still active. For this site, a typical water stress in beech trees was observed in the autumn, but no such stress existed in the understory species. Accordingly, the autumn broadband SR in the site showed active photosynthesis from the understory species, which was larger than that for only the tree canopy. In order to establish this fact, further detailed gas exchange measurements for the understory species are required and then up-scaling to canopy scale, together with Vc,25 tree canopy measurements. In other words, an alternative method is required by placing sensors close enough to the tree canopy to avoid any signals from nearby gaps. However, the asynchronous controls of physical (e.g., LAI) and physiological factors on canopy reflectance during the stress period in the site may also have been a principal factor for the autumn discrepancies.

[16] The small slope in the 900 m site was due primarily to high reflectance in optical infrared radiation. Compared with the other sites, this site was very windy and thus had high exposure of non-photosynthetic parts, which in turn triggered high reflectance in the optical infrared wavelengths and hence high broadband SR.

[17] Site-specific effects hindered drawing the general relationships between Vc,25 and broadband SR, though needed by gas exchange models. However, at least for the spring leaf-expansion period, it was not necessary to worry about such defect as R2 hit 0.60 (P < 0.000), when all data pooled together. What's more was that a statistically significant relationship was obtained when all data pooled together (Table 1), suggesting that a possible general regression model with acceptable error margin might be applicable to the models to replace current simulation methods of holding Vc,25 constant.

[18] Another type of broadband index, NDVI, showed a similarly significant relationship with Vc,25 (in the exponential model). However, broadband NDVI tended towards saturation at > 0.80, and ca. 68% of the broadband NDVI was > 0.80 in the growth period. In this sense, broadband NDVI was unable to track Vc,25 for most time periods, except at the start and end of the season, when low values prevailed. With regards seasonal patterns, broadband SR is superior over broadband NDVI in making linkages with Vc,25 seasonal trajectories.

[19] The findings may be criticized with respect to the impact of LAI on the relationship between broadband SR and Vc,25, especially when both the physical (LAI) and physiological aspects (photosynthetic capacity) apparently control seasonal courses of canopy reflectance, hence, vegetation indices in deciduous forests. Despite the fact that much focus has been placed on the physical factor, accumulating evidence in recent years [e.g., Wilson et al., 2000; Xu and Baldocchi, 2003; Wang et al., 2008] has indicated that there are large seasonal variations in photosynthetic activities that shape the pattern of canopy reflectance for deciduous forests. This is equally true for coniferous forests as the amplitude of seasonal variations in LAI does not correlate with that of canopy reflectance. Additional data at leaf-scale (in vivo reflectance spectra and simultaneous photosynthetic capacity from detached samples over the season), and at small FOVs (7.5° and 8°) canopy-scale (simultaneous in-situ canopy reflectance spectra at when and where detached samples were collected) further justified the findings in this study. Significant relationship between narrowband SR and Vc,25 was evident at both the leaf-scale (R2 = 0.66, P < 0.000), where there was no LAI effect on reflectance; and at the canopy-scale with small FOVs (R2 = 0.73, P < 0.000), where there were minimum effects from stand structure. The consistence in SR with Vc,25 at different scales suggested that the relationship found in the study was inherent.

[20] Remote sensing based vegetation indices may provide fast measurements on large spatial scales compared with tower-based broadband indices. However, remote sensing data have several inherent setbacks, by way of atmospheric effect, which reduce both the quality and frequency of the time series data, hence somehow inferior to tower-based data. The elimination of contaminated data points and subsequent interpolation of data gaps in remote sensing vegetation indices time series data require a sufficient priori knowledge and professional judgment, for which broadband indices may be used as possible calibration data. Future amalgamation of both data types can be highly promising and beneficial.

5. Conclusion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

[21] To accurately simulate canopy carbon fixation, seasonal variations in canopy photosynthetic capacity need to be incorporated into gas exchange models, which until now has largely been kept constant. As to whether this goal can be achieved fundamentally remains a question of the development of a sufficiently easy and fast method of canopy photosynthetic capacity estimation. However, parameterization of seasonal variations in photosynthetic capacity takes tremendous time and man-power, if solely based on traditional gas exchange measurements. In this study, we found tangible relationships between broadband SR (from meteorological sensor data) and Vc,25 – a major parameter in photosynthetic capacity calculation. Although the relationships may appear to be site-specific, overall broadband SR seasonal courses closely reflect Vc,25 seasonal trajectories. This, in itself, suggests the possibility of retrieving seasonal variations in Vc,25 from currently available meteorological data. Compared with its remote sensing counterpart, broadband SR is limited spatially, but is less sensitive to weather and thus capable of providing time series data for whole simulation periods. Currently several eddy covariance networks exit (e.g., AmeriFlux and EuroFlux) in which anchor stations readily provide routine broadband SR measurements. Consequently, our findings are on the verge of breaking new grounds in the estimation of seasonal variations in canopy photosynthetic capacity by providing a sufficiently easy and fast method. Notwithstanding however, further studies, especially for different forest ecosystems can be highly beneficial to this contemporary science of remote sensing.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information

[22] This research was supported by the JSPS Grant-in-Aid for Young Scientists (A) (Grant No. 18688007) awarded to Quan Wang.

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information
  • Chen, J. M., and J. Cihlar (1996), Retrieving leaf are index of boreal conifer forests using Landsat TM images, Remote Sens. Environ., 55, 153162.
  • Farquhar, G. D., S. von Caemmerer, and J. A. Berry (1980), A biochemical model of photosynthetic CO2 assimilation in leaves of C3 species, Planta, 149, 7890.
  • Harley, P. C., and J. D. Tenhunen (1991), Modeling the photosynthetic response of C3 leaves to environmental factors, in Modeling Crop Photosynthesis: From Biochemistry to Canopy, edited by K. J. Boote, and R. S. Loomis, pp. 1739, Crop Sci. Soc. of Am., Madison, Wis.
  • Huemmrich, K. F., T. A. Black, P. G. Jarvis, J. H. McCaughey, and F. G. Halls (1999), High temporal resolution NDVI phenology from micrometeorological radiation sensors, J. Geophys. Res., 104, 27,93527,944.
  • Ishizuka, K. (1974), Mountain vegetation, in The Flora and Vegetation of Japan, edited by N. Numata, pp. 173210, Kodansha, Tokyo.
  • Nogues, S., and L. Alegre (2002), An increase in water deficit has no impact on the photosynthetic capacity of field-grown Mediterranean plants, Funct. Plant Biol., 29, 621630.
  • Ross, J., and M. Sulev (2000), Sources of errors in measurements of PAR, Agric. Forest Meteorol., 100, 103125.
  • Tu, K. (2000), Modeling plant-soil-atmosphere CO2 exchange using optimality principles, Ph.D. dissertation, Univ. of N.H., Durham.
  • Wang, Q., J. Tenhunen, N.Q. Dinh, M. Reichstein, T. Vesala, and P. Keronen (2004), Similarities in ground- and satellite-based NDVI time series and their relationship to physiological activity of a Scots pine forest in Finland, Remote Sens. Environ., 93, 225237.
  • Wang, Q., A. Iio, J. Tenhunen, and Y. Kakubari (2008), Annual and seasonal variation in photosynthetic capacity of Fagus crenata along elevation gradients in the Naeba Mountains, Tree Physiol., 28, 277285.
  • Wilson, K. B., D. D. Baldocchi, and P. J. Hanson (2000), Spatial and seasonal variability of photosynthesis parameters and their relationship to leaf nitrogen in a deciduous forest, Tree Physiol., 20, 565587.
  • Wilson, K. B., D. D. Baldocchi, and P. J. Hanson (2001), Leaf age affects the seasonal pattern of photosynthetic capacity and net ecosystem exchange of carbon in a deciduous forest, Plant Cell Environ., 24, 571583.
  • Xu, L., and D. D. Baldocchi (2003), Seasonal trends in photosynthetic parameters and stomatal conductance of blue oak (Quercus douglasii) under prolonged summer drought and high temperature, Tree Physiol., 23, 865877.

Supporting Information

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Data and Methods
  5. 3. Results
  6. 4. Discussion
  7. 5. Conclusion
  8. Acknowledgments
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
grl24488-sup-0001-t01.txtplain text document0KTab-delimited Table 1.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.