JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 5 April 2007
Published in J Environ Qual 36:780-789 (2007)
DOI: 10.2134/jeq2005.0327
© 2007 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Tilley, D. R.
Right arrow Articles by Badrinarayanan, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Tilley, D. R.
Right arrow Articles by Badrinarayanan, H.
Agricola
Right arrow Articles by Tilley, D. R.
Right arrow Articles by Badrinarayanan, H.
Related Collections
Right arrow Wetlands and Aquatic Processes
Right arrow Ecosystem Management
Right arrow Field evaluation techniques
Right arrow Plant and Environment Interactions

TECHNICAL REPORTS

Wetlands and Aquatic Processes

Hyperspectral Reflectance Response of Freshwater Macrophytes to Salinity in a Brackish Subtropical Marsh

David R. Tilleya,*, Muneer Ahmedb, Ji Ho Sonc and Harish Badrinarayananb

a 1449 Agricultural Engineering Bldg. #142, Biological Resources Engineering Dep., Univ. of Maryland, College Park, MD 20742
b Dep. of Environmental Engineering, Texas A&M Univ.-Kingsville, Kingsville, TX 78363
c Division of Environmental Systems Engineering, Pukyong National Univ., Pusan 608-737, Korea

* Corresponding author (dtilley{at}umd.edu)

Received for publication August 25, 2005.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Coastal freshwater wetlands are threatened by increased salinity due to relative sea level rise and reduced freshwater inputs. Remote radiometric measurement of freshwater marsh canopies to detect small shifts in water column salinity would be useful for assessing salinity encroachment. We measured leaf hyperspectral (300–1100 nm) reflectance of freshwater macrophytes (cattail, Typha latifolia and sea oxeye, Borrichia frutescens) in a field study in a subtropical brackish (2.5–4.5 parts per thousand salinity, {per thousand}) marsh to determine salinity effects on visible and near-infrared spectral band reflectance and to identify reflectance indices sensitive to small (1{per thousand}) changes in wetland salinity. For sea oxeye, floating-position water band index [fWBI = R900/minimum(R930 – R980), where R{lambda} = reflectance at band {lambda}], normalized difference vegetation index [NDVI = (R774 – R681)/(R774 + R681)], and a proposed wetland salinity reflectance ratio (WSRR = R990/R933) were sensitive to salinity with R2 of 40, 35, and 65%, respectively (p < 0.01). For cattail, NDVI and photochemical reflectance index [PRI = (R531 – R570)/(R570 + R531)] were sensitive to salinity with R2 of 29 and 33%, respectively (p ≤ 0.01). Higher salinity significantly reduced mean reflectance of sea oxeye in 328- to 527-nm and 600- to 700-nm wavebands (p < 0.05), which corresponded to chlorophyll bands. Reflectance of cattail was not significantly affected by the highest salinity, although the spectral band most affected was 670 nm (p < 0.10), which is a chlorophyll a band. Our findings indicate that hyperspectral radiometry can detect the response of emergent freshwater plants to changes in wetland salinity, which would help with monitoring salinity effects on coastal wetlands.

Abbreviations: fWBI, floating-position water band index • NDVI, normalized difference vegetation index • NIR, near-infrared • PRI, photochemical reflectance index • SRVI, simple reflectance vegetation index • WSRR, wetland salinity reflectance ratio


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
THE water quality and ecological integrity of coastal wetlands is threatened by relative sea level rise and reduced flows of freshwater (Adam, 2002). Relative sea level rise increases the frequency and duration of inundation of coastal wetlands by brackish and saline seawater (Day and Templet, 1989; Nicholls et al., 1999), which contributes to the loss of freshwater marsh and reduces nutrient uptake, growth, and productivity of freshwater macrophytes (McKee and Mendelssohn, 1989; Baldwin and Mendelssohn, 1998). River dams, over withdrawal of aquifers, inter-basin water transfers, wetland drainage, and levees alter the quantity and timing of freshwater discharge to coastal ecosystems (Day and Templet, 1989; Rozema et al., 2000; Adam, 2002; Day et al., 2006). Understanding the geographic extent to which changes in relative sea level rise and altered freshwater discharge changes vegetation patterns, and composition and spatial distribution of coastal freshwater wetlands could be aided by radiometric remote sensing that can detect small changes in the salinity (i.e., one part per thousand salt, {per thousand}) of the wetland water column.

Evidence provided by Wang et al. (2002b) from a multispectral analysis of soil salinity effects on soybeans [Glycine max (L.) Merr.] determined that increased soil salinity reduced the near-infrared reflectance (NIR), but had no effect on visible reflectance. In a similar experiment with elephant grass (Pennisetum purpureum Schum.), Wang et al. (2002a) noted a significant reduction in NIR during the mid-growing season, but saw the effect disappear by harvest. In the same investigation of elephant grass, Wang et al. (2002a) saw a late-season increase in the 660-nm band at the highest salinity treatment. Reduced NIR from increased soil salinity has also been found for barley (Hordeum L.) and cotton (Gossypium L.) crops (Peñuelas et al., 1997b). In their soybean study, Wang et al. (2002b) identified a normalized difference vegetation index [NDVI = (R830 – R660)/(R830 + R660)] as significantly responsive to increased soil salinity. In a comprehensive review of the ability to use radiometric remote sensing to detect, map, and monitor salt-affected terrestrial ecosystems (e.g., cropland), Metternicht and Zinck (2003) pointed out that several investigators have found the NDVI sensitive to soil salinity. Although the NDVI is sensitive to soil salinity effects on crops, it also responds to other environmental factors such as nutrient levels, ground cover, and leaf area index (Strachan et al., 2002), which means that it can suggest the presence of salinity stress, but does not confirm it. Zhang et al. (1997) found that the NDVI was not significantly affected by salinity in a California (USA) salt marsh. They did, however, find that the soil adjusted vegetation index (SAVI) and the global environment monitoring index (GEMI) were slightly dependent on salt marsh salinity.

Although hyperspectral radiometry has been widely used to assess various biophysical and ecological properties of plants, crops, and ecosystems, such as quantifying the nutrient status of forests, marshes, and crops (Treitz and Howarth, 1999; Smith et al., 2002; Strachan et al., 2002; Smith et al., 2003; Townsend et al., 2003; Tilley et al., 2003; Tilley et al., 2005), classifying the trophic status of lakes and estuaries (Thiemann and Kaufmann, 2000; Koponen et al., 2002; Froidefond et al., 2002), evaluating soil salinity (Peñuelas et al., 1997b; Wang et al., 2002a, 2002b), and characterizing algal and red tide blooms (Kahru and Mitchell, 1998; Stumpf, 2001), its application to determine the effects of wetland water column salinity on the reflectance features of freshwater emergent macrophytes appears to have received little attention. There has been recent research that used hyperspectral reflectance to detect stress induced by low redox potentials in red maple (Acer rubrum) growing in a swamp (Anderson and Perry, 1996).

Therefore, there is a need to remotely sense salinity levels of coastal freshwater wetlands, such as tidal freshwater marshes, and there is evidence from terrestrial radiometric studies that salinity levels affect reflectance patterns. Given the need and promise of detecting wetland salinity levels with hyperspectral radiometry, our objectives in this study were (i) to characterize the effects of salinity on the reflectance of two freshwater macrophytes [cattail, Typha latifolia L. and sea oxeye, Borrichia frutescens (L.) DC] and (ii) to identify narrow spectral band reflectance indices that were sensitive to salinity levels.

Hyperspectral Reflectance Indices
The most basic reflectance indices in the visible and near-infrared (VNIR) spectrum are the direct measurements of reflectance in the blue, green, red, and near-infrared. The dominant leaf photopigments chlorophyll a and b absorb most strongly in the blue and red wavebands. Since photosynthesis is closely linked to chlorophyll levels and increased salinity reduces photosynthesis in at least some marsh plants (Ewing et al., 1995), substantial exposure to heightened salinity could eventually reduce the concentration of leaf chlorophyll, which could be detected as increased reflectance in the blue (e.g., R416) and red (e.g., R674) wavebands. Strachan et al. (2002) defined the Rshoulder as the average reflectance between 750 and 850 nm (a NIR waveband), which for healthy vegetation is the spectral region located between strong visible absorption and the water absorption feature centered at 960 nm. Ewing et al. (1995) found that salinity up to 28{per thousand} did not affect Rshoulder for Spartina patens, which is a salt marsh grass. However, Wang et al. (2002b) and Peñuelas et al. (1997b) found that soil salinity reduced near-infrared reflectance of important crops.

Demetriades-Shah et al. (1990) defined the red-edge (RE) as the wavelength of maximum slope at the red-near-infrared transition, and the dRE as the value of the first derivative of reflectance at the red-edge. Increased leaf chlorophyll content has been shown to increase the RE. If higher salinity negatively impacts leaf chlorophyll, then the RE and dRE should respond accordingly.

Combining individual spectral reflectance bands as simple ratio vegetation indices (SRVI) has been a common approach in remote sensing because it generally reduces the effects of spectral noise and allows for better temporal comparisons due to minimization of atmospheric effects (Carter and Miller, 1994). Commonly, SRVIs have consisted of the ratio of blue to red wavebands in an effort to detect responses due to changes in chlorophyll a and b concentrations. Tilley et al. (2003) found an SRVI (R493/R678) indicative of marsh ammonia levels based on cattail reflectance. Carter (1994) defined C420 as R420/R695 as an index for detecting plant stress caused by water and nutrient deficiencies.

The NDVI [(R774 – R681)/(R774 + R681)] was one of the earliest reflectance indices developed from multispectral imagery (Tucker, 1979). By measuring the relative difference between the chlorophyll sensitive red spectral band and a near-infrared band, the canopy scale NDVI indicates the greenness of the viewed landscape; that is, a higher NDVI indicates more live biomass or a higher leaf area index. One problem with using NDVI to estimate biomass is that it saturates in dense vegetation, which limits its applicability to indicate biomass levels. However, (Mutanga and Skidmore, 2004) suggested that a modified NDVI, which incorporated two narrow spectral NIR bands (i.e., 746 and 755 nm), improved the ability to predict biomass of a perennial grass (Cenchrus ciliaris). As mentioned above, Wang et al. (2002b) and other researchers (Metternicht and Zinck, 2003) have found that increased soil salinity reduced NDVI, but Zhang et al. (1997) found that NDVI did not change across a salinity gradient in a salt marsh. Normalized difference vegetation index is sensitive to several environmental factors in addition to salinity (Strachan et al., 2002), which means that it can suggest the presence of salinity stress, but cannot confirm it.

The floating-position water absorption band index (fWBI) was introduced by Strachan et al. (2002) as an index of water stress (i.e., matric stress) in corn. Development of the fWBI was based on the previous work of Peñuelas et al. (1997c) who demonstrated a strong relationship between a water index (R900/R970) and plant water concentration of Mediterranean vegetation. The fWBI is a modified simple reflectance ratio that measures the ‘depth’ of the water absorption band; i.e., it is the ratio of a near-infrared band located on the edge of the 900- to 1000-nm water absorption band and the spectral band of peak absorption (i.e., minimum reflectance) in the water absorption band. To our knowledge, the fWBI has not been related to salinity stress (i.e., osmotic stress); however, salinity and drought both reduce soil water potentials (Munns, 2002).

Increased salinity induces osmotic stress that reduces the ability of plants to take up water, which causes growth rates to decline, photosynthesis to slow, and stomatal conductance to decrease (DeLaune et al., 1987; Munns, 2002), which are similar to the effects of matric stress (i.e., water stress). In addition, osmotic stress and matric stress lead to a host of similar metabolic changes, like increased levels of abscisic acid (Bensen et al., 1988; He and Cramer, 1996). However, in addition to the osmotic stress that excess salt causes, plant species that cannot effectively store salt in vacuoles will build up toxic levels in leaves which will force premature senescence and reduced assimilate production (Munns, 2002). A plant water stress index based on leaf reflectance, like the fWBI, could be useful in detecting salinity stress in marshes, where matric stress can exist but is less common than in uplands. Shibayama et al. (1993) proposed a water stress index for rice paddies that was based on the first derivative of reflectance centered at 960 nm, which is similar to the fWBI because it depends on response in the 900 to 1000 nm water absorption feature. Given that osmotic stress and matric stress in plants share several effects (Munns, 2002) and that successful water stress reflectance indices have been developed based on the near-infrared water absorption feature, we explored whether an SRVI, defined as R990/R933 and called the wetland salinity reflectance ratio (WSRR), was responsive to salinity. The WSRR made use of attributes of the same 960 nm centered water absorption feature as the fWBI; however, the WSRR was the ratio of two narrow spectral bands located along opposing slopes of the feature, whereas the fWBI was the ratio of narrow spectral bands located at the minimum and maximum of the feature.

The photochemical reflectance index (PRI), defined as (R531 – R570)/(R531 + R570), has been found positively related to the rate of soil nutrient status (i.e., nitrogen, phosphorus, and potassium) for upland species (Gamon et al., 1997). To our knowledge it has not been identified as sensitive to salinity, but Thenot et al. (2002) found it responsive to water stress for a South American grain (Chenopodium quinoa) and a Mediterranean evergreen tree (Arbutus unedo). If leaf reflectance responded similarly to water and salinity stress, then PRI should respond to salinity stress. Peñuelas et al. (1997a) determined that the PRI of emergent wetland macrophytes was correlated to photosynthetic radiation-use efficiency, which was similar to non-wetland plants (Gamon et al., 1997). According to Gamon et al. (1997), the PRI includes R531 because of its sensitivity to non-radiative dissipation of photon energy via the xanthophyll cycle and the reference of R531 to R570 reportedly avoids overlapping spectral features unrelated to the xanthophyll signature under conditions of water stress. Previously, we have found the PRI to correlate with wetland ammonia availability (Tilley et al., 2003).


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
We conducted an observational study in an immature (2-yr-old), constructed marsh that served as the treatment filter in a recirculating shrimp farm and was undergoing vegetative self-organization (Tilley et al., 2002). We measured the in situ hyperspectral leaf reflectance of cattail and sea oxeye and the salinity, ammonia, nitrate, and total phosphorus levels of the wetland water column in which they were growing.

Site Description
A 2-yr-old 7.7-ha constructed wetland, located 10 km inland from Port Mansfield, Texas, USA (Fig. 1), which served as a recirculating treatment filter for a shrimp aquaculture facility (Loma Alta Shrimp Aquaculture Facility; 26°28'20'' N, 97°28'09'' W) was the study area. During the sampling period (July-August, 2000) water depth averaged 30.5 cm, but fluctuated between 15 and 45 cm. Total precipitation during the 5-wk sampling period was 170 mm (NOAA, 2002). Wind speeds were not measured at the site, but the semiarid coastal region experiences average summer winds that range between 6 and 7 m s–1 (NOAA, 2002). The wetland was excavated down to the Incell Clay soil layer (NRCS, 2006). The physical layout, planting scheme, second year plant survival, management, and sampling station locations were previously described (Tilley et al., 2002).


Figure 1
View larger version (28K):
[in this window]
[in a new window]

  		Fig. 1. Geographical location and layout of brackish constructed marsh with location of sampling stations within marsh.

 
By the second year of operation (the year of this study) the wetland had undergone self-design whereby cattail became the overwhelmingly dominant plant species, occupying nearly two-thirds of the wetland area. Cattail occurs naturally in the area surrounding the shrimp farm in coastal freshwater marshes that are periodically inundated with seawater from storm surges (Britton and Morton, 1997). Sea oxeye was mainly found along the fringe of the study wetland. Sea oxeye occurs naturally along the Texas coast behind low beach ridges in salty soils, which are areas infrequently inundated (Britton and Morton, 1997). During the study period both species were green and apparently vibrant with no overt indications of stress. In subsequent years, which were not included in this study, the shrimp farm managers allowed the salinity of the marsh to reach 10 to 15{per thousand} for extended periods (i.e., weeks) and both species appeared less green with more brown leaf surface area. Sampling stations were located where sea oxeye and cattail were adjacent to each other.

Salinity and Reflectance Measurements
Salinity of the wetland was measured in the morning at mid-depth with a conductivity meter (Extech Oyster, Inc., Tampa, FL) 1 to 2 h before hyperspectral reflectance measurements. Simultaneous to measuring conductivity, water samples were collected from the wetland at mid-depth, preserved, and transported to an analytical lab for nutrient analysis. Nutrients analyzed included ammonium, nitrate, and total phosphorus and were determined using USEPA approved methods for the Hach DR2000 Spectrophotometer (Hach, 1991; APHA, 1992). More details concerning wetland water quality dynamics can be found in Tilley et al. (2002).

Spectral reflectance of cattail and sea oxeye were collected weekly during a 5-wk period that extended from 31 July to 28 Aug. 2000. Four sampling stations were established from the wetland inlet to the outlet along the main flow path where salinity fluctuated regularly (Fig. 1). A field portable spectroradiometer (ASD FieldSpec UV/NIR-Open Sky 300–1100 nm, Analytical Spectral Devices, Inc., Boulder, CO) was used to measure leaf reflectance in 1.635-nm intervals between 1000 and 1200 h. Skies for sampling dates were either clear or had sparse cloud cover. Before collecting each leaf reflectance measurement, the reflectance of a white Spectralon panel (Labsphere, Inc., North Sutton, NH) was taken. The RS3 software (ASD, 2003) automatically calculated percentage plant reflectance by dividing plant sample reflectance by reflectance of the white panel. Samples were taken with the 25 degree field-of-view bare fiber (i.e., no foreoptic) held 1 to 2 cm from the leaf surface, which corresponded to a leaf sample area of 0.68 to 2.7 cm2. The spectroradiometer was set to collect five subsamples with intercollection intervals of 1 s and an integration time of 17 ms. The mean of the five subsamples was used during subsequent correlation and regression analyses.

Data Analysis
Effects of salinity on the spectral reflectance of each species was determined based on the different response between low and high salinity data groups. The low salinity group (LO) had salinities <3.5{per thousand}, while the high salinity group (HI) had salinities above 3.5{per thousand}. A salinity of 3.5{per thousand} was chosen because it was the middle of the salinity range (2.5–4.5{per thousand}) observed during the study period.

Salinity spectral correlations (i.e., Pearson's coefficient for correlation between narrow spectral band reflectance and marsh salinity as a function of wavelength) were developed for each species to determine the broad relationships between marsh salinity and visible and NIR. The spectral correlations provided insight into which spectral bands should be combined as simple reflectance ratios for testing as useful indicators of wetland salinity, which was similar to an analytical method used by Read et al. (2002). Simple reflectance ratios reduce the noise of the spectral data, provide a single metric for cross comparison of spectra collected under different environmental conditions, and are related to environmental and plant variables (Haboudane et al., 2004). Simple reflectance ratios can relate sensitive spectral bands to one another or relate a sensitive band to a non-sensitive one. Finally, we used SPSS for Windows 10.0.5 (SPSS, 1999) for linear regression, correlation, and ANOVA. Reflectance indices tested for sensitivity to marsh water column salinity are given in Table 1. Level of significance for correlations, regressions, and ANOVA was taken as a p value <10% (i.e., p < 0.10).


View this table:
[in this window]
[in a new window]

  		Table 1. Formulation of reflectance indices tested in linear regression analysis for response to salinity.{dagger}

 

   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Wetland Salinity and Water Quality
The LO salinity group had a mean salinity of 2.8{per thousand} with a range of 2.5 to 3.5{per thousand}, while the HI salinity group had a mean salinity of 4.0{per thousand} with a range of 3.5 to 4.5{per thousand} (Table 2). Means were significantly different (p < 0.10). Nutrient levels, which can also affect reflectance, varied during the experiment. Total phosphorus and total ammonia ranged from 0.01 to 0.86 mg P L–1 and from 0.92 to 1.71 mg N L–1, respectively, while nitrate was consistently well below 1.0 mg N L–1 (Tilley et al., 2002). The studied wetland was designed to reduce phosphorus, nitrogen, biological oxygen demand (BOD5), and suspended sediment levels from the effluent of an 8-ha shrimp farm so water could be recycled back to the shrimp ponds during the shrimp growing season. The wetland worked according to design, reducing the concentration of total phosphorus and suspended sediment along the direction of flow and maintaining low ammonia, nitrate, and BOD5 concentrations (Tilley et al., 2002).


View this table:
[in this window]
[in a new window]

  		Table 2. Salinity characteristics of low (LO) and high (HI) salinity groups.

 
Species Differences
Cattail, the more dominant wetland cover, reflected 2 to 6% less than sea oxeye across the visible spectrum (400–700 nm, p < 0.10) but generally reflected 10% more in the near-infrared (700–1100 nm, p < 0.10, Fig. 2). The average leaf NDVI for cattail was 0.82, vs. 0.65 for sea oxeye. Cattail also exhibited an average RE of 718 nm that was 15 nm greater than that of sea oxeye.


Figure 2
View larger version (22K):
[in this window]
[in a new window]

  		Fig. 2. Mean reflectance of cattail (Typha latifolia L.) and sea oxeye [Borrichia frutescens (L.) DC] over 31 July to 28 Aug. 2000 sampling period and the probability (p value) of not being different at each spectral band.

 
Plant Response to Salinity
For sea oxeye the HI salinity group had a higher mean reflectance across the visible and near-infrared spectrum than the LO salinity group (Fig. 3a). However, only the visible had significantly different spectral bands (p < 0.10, 328–501 nm and 660–690 nm). Cattail, on the other hand, did not exhibit any significantly different spectral bands due to salinity (Fig. 3b).


Figure 3
Figure 3
View larger version (47K):
[in this window]
[in a new window]

  		Fig. 3. Mean leaf reflectance spectra for the high (HI) and low (LO) salinity groups for (a) sea oxeye and (b) cattail with the probability (p-value) that spectral bands were not different.

 
Salinity spectral correlations (Fig. 4a) showed that sea oxeye exhibited positive correlations with salinity (r = 0.40 and 0.45, respectively) in the chlorophyll a and b absorption bands (p < 0.10), but had no detectable effect on green or near-infrared wavebands. Sea oxeye's positive correlation between the reflectance of chlorophyll sensitive bands (i.e., blue and red) and salinity was similar to the response Wang et al. (2002b) measured for soy bean. The spectral correlations indicated that sea oxeye's ability to absorb photosynthetically active radiation was impaired by higher salinity. The correlation of blue and red reflectance to salinity for cattail was not statistically significant (Fig. 4b, p >> 0.10). The differential species response to salinity may indicate differing physiological abilities to tolerate brackish water.


Figure 4
View larger version (26K):
[in this window]
[in a new window]

  		Fig. 4. Salinity spectral correlation for (a) sea oxeye and (b) cattail with probability (p value) that correlation coefficient was not equal to zero.

 
Our observation of a negative correlation between NIR and salinity for cattail was not significant at p < 0.10, but was at p < 0.20 (Fig. 4b). Studies of agriculturally important crops have demonstrated a positive relationship between NIR and salinity (Wang et al., 2002b). Cattail has been shown to tolerate salinities up to 10{per thousand} (Konisky and Burdick, 2004), which is a concentration 2 to 4 times higher than what we observed. At the comparatively moderate salinities observed in our study, cattail appears to reflect less NIR (i.e., absorb more), which may be caused by a physiological mechanism similar to mangroves or other halophytes that allows the absorption of additional solar power (especially in the NIR) to separate freshwater from salt (Odum, 1984; McClanahan and Odum, 1991). An alternative explanation is that the eutrophic conditions of the treatment marsh created a higher demand for carbon dioxide and photosynthetically active radiation (Fig. 2), which increased leaf gas exchange and leaf heat load. The higher gas exchange and heat load increased water demand, which was met by using more near-infrared radiation to separate freshwater from salt.

Response of Reflectance Indices to Salinity
Sea oxeye had seven of eleven leaf reflectance indices that were responsive to marsh salinity (Table 3). Our newly proposed WSRR had the strongest correlation to salinity of all indices tested (r2 = 0.65, p < 0.01), while fWBI was the second strongest and NDVI was third. Cattail, on the other hand, had only two of eleven indices respond to salinity. These two indices included the PRI, which was the most responsive to salinity (r2 = 0.33, p < 0.01), and the NDVI, which was a close second (r2 = 0.29, p = 0.01).


View this table:
[in this window]
[in a new window]

  		Table 3. Coefficient of determination (r2), sign of slope (+, –, or not significant [ns] at p < 0.10), significance level (p value), and root mean square error (RMSE) of salinity ({per thousand}) for linear regression between reflectance index and marsh salinity.{dagger}

 
Since reflectance indices can be affected by various environmental factors, especially nitrogen availability, we tested whether nutrient levels (phosphorus, ammonia, and nitrate) within the wetland water column were correlated to salinity and whether any of the reflectance indices were correlated to any of the nutrients. Salinity was not significantly correlated to phosphorus, ammonia, or nitrate. In addition, for sea oxeye none of the reflectance indices that we found significantly correlated to salinity (i.e., R416, R674, C420, SRVI, NDVI, fWBI, WSRR) were significantly correlated to any of the nutrients. For cattail, on the other hand, which in general exhibited less responsiveness to salinity, we found that its PRI was also significantly correlated to ammonia (r2 = 0.23, p < 0.05), but the relationship was opposite to the salinity effect. In a stepwise regression for the PRI of cattail with salinity and ammonia as candidate independent variables, ammonia was removed from the model in favor of salinity at p < 0.05, indicating that PRI was more responsive to salinity than ammonia. The counteractive effects of nitrogen and salinity on cattail's PRI suggested that PRI was likely a poor indicator of marsh salinity. Photochemical reflectance index was the only reflectance index we found to be affected by ammonia or any other water column nutrient, which allayed our concerns that factors other than salinity were affecting the reflectance indices.

Salinity reduced the NDVI of sea oxeye at double the rate it did for cattail (Fig. 5). For sea oxeye, each one part per thousand increase in salinity over the observed range reduced the NDVI by 0.10, while for cattail, each part per thousand increase in salinity decreased the NDVI by 0.05. Salinity explained 35% of the variation in NDVI for sea oxeye (Fig. 5a), and explained slightly less (29%) for cattail (Fig. 5b). This indicated that NDVI of marsh leaves can be an indicator of salinity stress, but that species differences are an important consideration.


Figure 5
View larger version (13K):
[in this window]
[in a new window]

  		Fig. 5. Normalized difference vegetation index (NDVI) of (a) sea oxeye (p < 0.01) and (b) cattail (p = 0.01) as a function of salinity.

 
We observed that the PRI, which other authors have found indicative of plant stress (Peñuelas et al., 1994), was reduced by salinity for cattail (Fig. 6). Each one part per thousand increase in salinity reduced cattail's PRI by 0.0155 (Fig. 6). For sea oxeye, regression of the PRI of individual samples to salinity was not significant (Table 3), however when we grouped individual samples by salinity level, the mean PRI of each salinity level was significantly affected by salinity (r2 = 0.58, Fig. 7). Making a similar adjustment to cattail's PRI data increased the explanatory power of salinity (r2 = 0.52), but at a reduced level of significance (p = 0.11, Fig. 7).


Figure 6
View larger version (13K):
[in this window]
[in a new window]

  		Fig. 6. Photochemical reflectance index (PRI) of (a) sea oxeye (p > 0.10) and (b) cattail (p = 0.01) as function of wetland salinity.

 

Figure 7
View larger version (15K):
[in this window]
[in a new window]

  		Fig. 7. Mean photochemical reflectance index (PRI) per salinity level as a function of salinity for sea oxeye (p < 0.10) and cattail (p = 0.11).

 
Although the PRI and fWBI have been found indicative of water stress for crops and evergreen trees (Thenot et al., 2002; Strachan et al., 2002), this is likely not the reason why they both varied during our study because both cattail and sea oxeye were continuously flooded during the entire experiment. However, continuous flooding can have negative effects on marsh plant respiration (Naidoo et al., 1992), which may ultimately affect the PRI and fWBI. Rather than responding to water stress, the PRI and fWBI may have been influenced by the stress of elevated salinity, which has many physiological effects similar to drought (Munns, 2002). We were unaware of other researchers finding the PRI or fWBI as indicative of salinity stress.

Since low nitrogen availability can reduce marsh plant photosynthesis (Tilley et al., 2005) and high salinity lowers photosynthesis (Munns, 2002), it is likely that low nitrogen may exacerbate the stress of high salinity for freshwater macrophytes. We wondered whether the fWBI response to salinity (Fig. 8) was confounded by ammonia levels in the marsh. To test this notion, we regressed fWBI against salinity using only the portion of the data that had a total ammonia concentration greater than 1.1 mg N l–1 (Fig. 9). This improved the ability of salinity to explain variability in fWBI, and indicated that there was an interaction between salinity and nitrogen in affecting leaf fWBI.


Figure 8
View larger version (13K):
[in this window]
[in a new window]

  		Fig. 8. Floating-position water band index (fWBI) as a function of salinity for (a) sea oxeye (p < 0.01) and (b) cattail (p = 0.26).

 

Figure 9
View larger version (11K):
[in this window]
[in a new window]

  		Fig. 9. Floating-position water band index (fWBI) of sea oxeye as function of salinity for samples with ammonia >1.1 mg L–1 (p < 0.01).

 
Of all the reflectance indices we tested for sea oxeye, our new simple reflectance ratio, the WSRR, had the most variation explained by salinity (r2 = 0.65, Fig. 10a). However, the WSRR of cattail did not respond to salinity, which may have been because cattail was not stressed at the observed salinities (Konisky and Burdick, 2004). In general, the reflectance indices of sea oxeye were more sensitive to salinity than cattail over the observed range (2.5 to 4.5{per thousand}), indicating that cattail had a higher tolerance to salinity than sea oxeye.


Figure 10
View larger version (13K):
[in this window]
[in a new window]

  		Fig. 10. The wetland salinity reflectance ratio (WSRR = R990/R933) as a function of salinity for (a) sea oxeye (p < 0.0001) and (b) cattail (p = 0.45).

 
Since our study was only observational and not manipulative, the precision with which the WSRR can indicate salinity stress of emergent freshwater wetland plants needs to be more fully investigated. A first step would be to conduct a designed greenhouse, plant-scale experiment with various salinity and nitrogen treatments and wetland species to more convincingly show that the WSRR could be a strong radiometric index for detecting salinity stress in marsh plants. If the WSRR were shown to work for a variety of marsh species across moderately large salinity and nitrogen ranges, then the question of scaling the leaf-based index (i.e., WSRR) to canopy level would need to be addressed. As Huang et al. (2004) highlighted, it is not trivial to extend the utility of a leaf-based reflectance index to entire canopies because of canopy geometry, ground cover, and atmospheric effects. For example, hyperspectral imaging data collected by the airborne visible/infrared imaging spectrometer (AVIRIS) must be corrected for atmospheric water vapor absorption using a model such as ATREM (Gao et al., 1993), which could strongly suppress the signal located within a water vapor absorption feature, such as WSRR. However, if the WSRR were scalable to the canopy, and atmospheric effects were overcome without great signal loss, then it could be used as a remote sensing tool for monitoring the salinity stress of tidal freshwater marshes at a meaningful geographic scale. This would provide the ability to monitor changes in the location and extent of the landward front of the freshwater/seawater interface.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
In summary, an observational field experiment in a brackish (2.5–4.5{per thousand}) constructed marsh determined that salinity increased the leaf reflectance of sea oxeye, an emergent freshwater macrophyte, in blue and red wavebands, which are major absorption bands of chlorophyll a and b, but that salinity at these brackish levels did not significantly affect the reflectance of cattail. However, the 667-nm spectral band, which is within the chlorophyll absorption waveband, was the band most significantly (i.e., lowest p value) affected by salinity for cattail. Near-infrared reflectance was only weakly affected by salinity for cattail. The different hyperspectral reflectance patterns between the species appears to be attributable to their different tolerances and responses to salinity. Three previously developed reflectance indices (i.e., NDVI, PRI, and fWBI) and one new simple reflectance ratio (i.e., wetland salinity reflectance ratio, WSRR = R990/R933) were found to be sensitive to marsh salinity, although not always for both species. For sea oxeye, NDVI, fWBI, and WSRR were significantly altered by salinity, while for cattail only NDVI and PRI were significantly affected. However, the PRI of cattail was also affected by observed wetland ammonia levels in the opposite direction. As an observational study, there was no manipulation of salinity levels. Since cattail can reportedly withstand salinities up to 10{per thousand} before major changes in biological processes takes place (Konisky and Burdick, 2004), the salinities observed in the studied marsh are well below the level at which cattail reflectance would exhibit a major response. This study provides the foundation needed to pursue further development of radiometric models that are predictive of low level salinity changes in coastal freshwater wetlands.


   ACKNOWLEDGMENTS
 
The H.P. El Sauz Ranch graciously allowed access to the Loma Alta Shrimp Aquaculture Facility and provided partial funding. Other funding came from the U.S. Department of Energy HBCU/MI–Environmental Technology and Waste Management program. The Maryland Agricultural Experiment Station, College Park provided partial financial support. Three anonymous reviewers provided comments that improved the final version of this manuscript.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Tilley, D. R.
Right arrow Articles by Badrinarayanan, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Tilley, D. R.
Right arrow Articles by Badrinarayanan, H.
Agricola
Right arrow Articles by Tilley, D. R.
Right arrow Articles by Badrinarayanan, H.
Related Collections
Right arrow Wetlands and Aquatic Processes
Right arrow Ecosystem Management
Right arrow Field evaluation techniques
Right arrow Plant and Environment Interactions

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Soil Science Society of America Journal Journal of Plant Registrations The Plant Genome