Pugnaire F.I. Valladares F. Functional Plant Ecology

Подождите немного. Документ загружается.

Spatial scale

Large Long

ShortSmall

Temporal scale

FIGURE 22.2 Interdependent spatial and temporal scales covered by remote sensing. At coarser

geographic scales (large pixel sizes, grain sizes, or IFOV), processes tend to be detectable over larger

time frames, whereas many processes at fine spatial scale (small pixel sizes, grain sizes, or IFOV) can

often be detected over short time periods.

FIGURE 22.3 (See color insert following page 684.) True-color AVIRIS image for a region of Ventura

County in southern California showing the Pacific Ocean (bottom), the western edge of the Santa Monica

Mountains (center right), and developed and agricultural areas (top left). The different colored bands

receding into the background illustrate the spectral dimension (Z dimension), in this case 224 different

spectral bands (wavelengths), ranging from near-ultraviolet bands (foreground) to the infrared bands

(background). Much of the information content of this image is present in this spectral dimension.

Francisco Pugnaire/Functional Plant Ecology 7488_C022 Final Proof page 660 18.4.2007 11:51am Compositor Name: VAmoudavally

660 Functional Plant Ecology

spectral bands, even at larger spatial scales. Sensor look angle and radiation polari-

zation provide additional sampling dimensions that may be relevant in specific cases

(Barnsley 1994). For example, certain structural features of vegetation are sensitive to

polarization of optical radiation (Vanderbilt et al. 1990) or microwave radiation (Hess and

Melack 1994).

VEGETATION INDICES

In terrestrial systems, vegetated and nonvegetated areas can be readily distinguished based on

their contrasting reflectance patterns in the red and near-infrared (NIR) spectral regions

(Figure 22.4). This contrast results from differential pigment absorption in the red and NIR

wavebands, and illustrates the radiation requirements for photosynthesis. Photosynthetic

pigments (primarily chlorophylls and carotenoids) absorb most effectively in the visible

region (400–700 nm), in which energy is most abundant and strong enough to drive electron

transport, yet weak enough to avoid excessive damage to biological molecules. By contrast,

there is insufficient energy in the NIR to drive photosynthesis; consequently, photosynthetic

pigments cannot use or absorb these wavelengths, and vegetation canopies effectively scatter

(reflect and transmit) most NIR radiation.

Numerous vegetation indices have been developed that characterize the contrasting

reflectance on either side of the red edge at 700 nm, including the normalized difference

vegetation index (NDVI), the simple ratio (SR), the soil-adjusted vegetation index and its

derivatives (SAVI, TSAVI, SARVI), the greenness vegetation index (GVI), the perpendicular

vegetation index (PVI), and the enhanced vegetation index (EVI) (Perry and Lautenschlager

1984, Baret and Guyot 1991, Wiegand et al. 1991, Huete et al. 1997, 2002). Other indices are

based on the slope or inflection point of the red edge (Curran et al. 1990, Johnson et al. 1994,

Gitelson et al. 1996). Essentially, all of these indices collapse the full spectrum into a single,

readily usable value that scales with green canopy cover, absorbed radiation, leaf area index

0.4

0.3

0.2

0.1

0.0

Blue

400 500 600 700

Wavelength (nm)

Reflectance

800 900 1000

Red

Soil

Near-IR

Vegetation

FIGURE 22.4 Typical reflectance spectra for green vegetation and bare soil in the visible (400–700 nm)

to NIR (>700 nm) region. Chlorophyll absorbs well in the visible (particularly the blue and red regions),

but not in the infrared, resulting in the characteristic vegetation red edge at 700 nm. This red edge

feature is generally missing in nonphotosynthetic surfaces. The contrasting red and NIR reflectance is

the basis for many vegetation indices that depict relative amounts of green vegetation cover. Note also

the slight water absorption feature (dip) between 900 and 1000 nm, which can serve as an indicator of

vegetation water content.

Francisco Pugnaire/Functional Plant Ecology 7488_C022 Final Proof page 661 18.4.2007 11:51am Compositor Name: VAmoudavally

Ecological Applications of Remote Sensing at Multiple Scales 661

(LAI), or related measures of vegetation abundance. Of all these indices, the most commonly

used index is undoubtedly the NDVI:

NDVI ¼

NIR

 R

RED

NIR

þ R

RED

, (22:1)

where R

NIR

is the reflectance in the NIR region and R

RED

is the reflectance (or radiance) in the

red region (Figure 22.4).

Each of these vegetation indices suffers from certain well-discussed limitations (Holben

1986, Myneni and Asrar 1994, Myneni and Williams 1994, Sellers et al. 1996b). Many of these

limitations arise because reflectance of any spectral region is necessarily influenced by

multiple factors that can confound a simple interpretation of a given reflectance signature.

Complex landscapes contain many components that can have multiple, overlapping effects on

any single index, and the concept of a pure index becomes elusive, particularly at large spatial

scales, where many landscape components contribute to the measured signal. Additional

errors arise from atmospheric effects (absorption or scattering), variation in solar angle,

and sensor calibration drifts. Added to this is the fact that each sensor has unique spectral

characteristics (spectral response and bandwidth). For example, the red and NIR bands vary

in width and position across sensors, meaning that, in reality, many NDVI formulations exist,

leading to problems when comparing data across sensors. Consequently, it is often necessary

to correct indices for these confounding factors, and this is now a common practice with

global NDVI datasets (Goward et al. 1994, James and Kalluri 1994, Los et al. 1994, Sellers

et al. 1994, 1996b, Townshend 1994). Despite these flaws, all of these vegetation greenness

indices provide some measure of light absorption by green vegetation, and thus can be used to

distinguish relative levels of energy capture for photosynthetic processes. This fundamental

relationship is the basis for most models of photosynthetic carbon uptake driven by remote

sensing, as further discussed later.

Some of the most potent and spectacular applications of remote sensing in ecology are

derived from the ability to detect spectral features associated with specific, biologically

important compounds (Table 22.2). Some of these compounds, notably photosynthetic

pigments (chlorophylls and carotenoids) and photoprotective pigments (carotenoids and

anthocyanins), are positioned for capturing solar energy; consequently, they are well-suited

for detection from above. Because photosynthetic pigments control light absorption for

photosynthetic carbon uptake, their detection can be linked to photosynthetic production

models.

NDVI provides a good example of an operational index that is influenced by both

vegetation structure and chlorophyll content (Figure 22.4), thus providing a powerful measure

of vegetation greenness and radiation absorption. Over terrestrial regions, global NDVI

dynamics are significantly correlated with seasonal and latitudinal variation of atmospheric

content (Tucker et al. 1986, Fung et al. 1987), demonstrating a strong influence of

terrestrial photosynthesis on atmospheric CO

levels. NDVI is also strongly linked to primary

production at continental scales (Goward et al. 1985). This realization of strong links between

NDVI, CO

fluxes, and primary production, along with the increasing availability of stand-

ardized global NDVI datasets, has spurred tremendous progress in global studies of carbon

flux, as further discussed later.

NEW OPPORTUNITIES WITH HYPERSPECTRAL SENSORS

The advent of hyperspectral sensors—those with many, narrow, adjacent bands—is

now allowing exploration of many additional biologically active compounds with spectral

Francisco Pugnaire/Functional Plant Ecology 7488_C022 Final Proof page 662 18.4.2007 11:51am Compositor Name: VAmoudavally

662 Functional Plant Ecology

reflectance. Operational methods for reliably quantifying many of these compounds are

now available (Table 22.2), and some have been discussed in recent reviews (Curran 1989,

Wessman 1990, Pen

uelas and Filella 1998). A number of pigments, notably carotenoids

(accessory photosynthetic pigments; Young and Britton 1993) and anthocyanins (phenolic

pigments; Strack 1997), may serve as indicators of stress or leaf senescence in many cases.

Additionally, distinct absorption features for water vapor and liquid water are now remotely

detectable (Gao and Goetz 1990, Green et al. 1991, 1993, Roberts et al. 1997, Gamon et al.

1998, Ustin et al. 1998, Serrano et al. 2000, Sims and Gamon 2003). Water vapor features

are now used to remove confounding atmospheric effects in the derivation of apparent

surface reflectance from uncorrected radiance signals (Green et al. 1991, 1993, Roberts et al.

1997). Liquid water absorption features can provide useful indices of LAI and vegetation

cover (Roberts et al. 1997, Gamon et al. 1998, Ustin et al. 1998), canopy moisture status

(Penuelas et al. 1993, 1997b, Zhang et al. 1997, Gamon et al. 1998, Ustin et al. 1998), and

evapotranspiration (Claudio et al. 2006).

In addition to pigments or water status, absorption features of other biochemically

relevant compounds, including lignin or nitrogen, can be detected with hyperspectral sensing,

leading to new applications in the detection and mapping of vegetation functional state.

Remote detection of lignin or nitrogen content has sometimes provided inputs for models of

ecosystem production and N cycling, and successional change (Wessman et al. 1988,

TABLE 22.2

Examples of Biologically Important Compounds Detectable with Spectral Reflectance

Compounds Function References

Chlorophylls Photosynthesis (PAR absorption) Gitelson and Merzlyak (1994, 1996, 1997),

Gitelson et al. (1996), Carter (1994),

Johnson et al. (1994), Pen

uelas et al.

(1994, 1995a), Gamon et al. (1995),

Yoder and Pettigrew-Crosby (1995)

Carotenoids

(carotenes and

xanthophylls)

Photosynthesis (PAR absorption)

and protection (photoprotection,

antioxidants)

Gamon et al. (1990, 1992, 1997),

Pen

uelas et al. (1994, 1995a, 1997a),

Filella et al. (1996), Gamon and Surfus (1999),

Gitelson et al. (2002), Sims and Gamon (2002)

Flavonoids

(anthocyanins)

Protection (antioxidants,

pathogen deterrent, sunscreen)

Gould et al. (1995), Coley and Barone (1996),

Gamon and Surfus (1999), Gitelson et al. (2001)

Water Essential for structural support

and most metabolic processes

Bull (1991), Carter (1991),

Pen

uelas et al. (1993, 1994, 1997b),

Gao and Goetz (1995), Roberts et al. (1997),

Zhang et al. (1997), Gamon et al. (1998),

Ustin et al. (1998), Serrano et al. (2000),

Sims and Gamon (2003), Claudio et al. (2006)

Lignin Plant cell wall structure, resists

decomposition

Peterson et al. (1998), Wessman et al. (1988),

Johnson et al. (1994),

Gastellu-Etchegorry et al. (1995)

Nitrogen-containing

compounds

(e.g., proteins and

pigments)

Needed for many

metabolic processes

Peterson et al. (1988), Pen

uelas et al. (1994),

Filella et al. (1995) Gamon et al. (1995),

Gastellu-Etchegorry et al. (1995),

Johnson et al. (1994),

Yoder and Pettigrew-Crosby (1995),

Asner and Vitousek (2005)

Cellulose Plant cell wall structure Gastellu-Etchegorry et al. (1995),

Martin and Aber (1997)

Francisco Pugnaire/Functional Plant Ecology 7488_C022 Final Proof page 663 18.4.2007 11:51am Compositor Name: VAmoudavally

Ecological Applications of Remote Sensing at Multiple Scales 663

Aber et al. 1990, Matson et al. 1994, Martin and Aber 1997, Peterson et al. 1988, Asner and

Vitousek 2005; however, see cautions by Curran 1989, Curran and Kupiec 1995, Grossman

et al. 1996).

The emergence of hyperspectral sensors is also spurring the application of new analytical

procedures that promise to add insight into ecological questions. In contrast to simple indices,

which are typically limited to information from one or two broad spectral bands, these

methods often take advantage of the additional information present in the shape of narrow-

band spectra, and are often able to resolve subtle features not detectable with broad-band

indices. These alternative approaches include derivative spectra (Demetriades-Shah et al.

1990), continuum removal (Clark and Roush 1984), hierarchical foreground or background

analysis (Pinzon et al. 1998), and a variety of other feature fitting methods (e.g., Gao and

Goetz 1994, 1995). Another promising example is provided by spectral mixture analysis,

which models a spectrum as a mix of component endmember spectra (Roberts et al. 1993,

1997, Ustin et al. 1993). These endmembers can be spectra of pure components (e.g., species,

canopy components, minerals, or soil types; Roberts et al. 1993, 1998), or can be spectra

derived from recognizable landscape components present in the image itself (Tompkins et al.

1997). Because spectral mixture analysis uses the additional leverage of multiple spectral

bands (instead of the few bands used in most spectral indices), it can be less sensitive to

confounding factors, assuming all of the appropriate spectral endmembers are properly

included in the model. Because it unmixes a spectrum to fundamental components (e.g.,

areas of bare ground and vegetation), it can be applied to situations in which components

are smaller than the spatial resolution of the detector (pixel size). On the other hand, spectral

mixture analysis is computationally intensive, and the resulting products can be highly

dependent on the particular endmembers selected (Roberts et al. 1998), making quanti-

tative interpretations difficult. Although many challenges to interpretation and validation

of hyperspectral data remain, the continued emergence of hyperspectral sensors, combined

with improved computational capabilities, undoubtedly leads to increased refinement and

application of new analytical methods applicable to ecological studies.

REMOTE SENSING AS A FUNCTIONAL MAPPING TOOL

The traditional strength of remote sensing lies in its ability to make objective maps of regions

much larger than can be sampled from the ground. When image features are linked to objects

or events on the surface, these maps allow us to greatly extend our view of ecologically

significant patterns and processes. Historically, aerial photography has provided a useful tool

for local- to regional-scale mapping, and this application has been well described elsewhere

(Knipling 1969, Yost and Wenderoth 1969, Salerno 1976, Lillesand and Kiefer 1987). In

contrast to the high spatial resolution (fine detail) of aerial photography, many current

imaging sensors operate at coarser spatial resolution (typically, a pixel size ranging from

several meters to a thousand meters in diameter), and this resolution typically degrades with

sampling distance (Figure 22.2). However, current imaging spectrometers offer added benefits

of a digital data format and supplementary spectral information that can be thought of as

multiple data layers for a given image (Figure 22.3). Multitemporal coverage, now provided

by many satellite and some aircraft sensors (Table 22.1), adds an additional dimension that is

particularly useful in examining time-dependent processes, including phenological, succes-

sional, and land-use change (Justice et al. 1985, Malingreau et al. 1989, Hobbs 1990, Hall et al.

1991, DeFries and Townshend 1994a,b). Exploration of spectral and temporal information in

a spatial context offers powerful new ways of distinguishing features and functional states.

This combination of spatial, temporal, and spectral dimensions in a structured, multidimen-

sional data volume can readily reveal ecologically significant patterns and processes.

Francisco Pugnaire/Functional Plant Ecology 7488_C022 Final Proof page 664 18.4.2007 11:51am Compositor Name: VAmoudavally

664 Functional Plant Ecology

The uniform data format provides potent inputs for digital mapping and modeling tools

(e.g., geographic information systems; Jensen 1996) that are now revolutionizing the field of

ecology. Landscape ecology (Turner and Gardner 1991) and global ecology (Gates 1993,

Solomon and Shugart 1993, Walker and Steffen 1996) provide examples of new fields

emerging from this synthesis of pattern and process.

A simple illustration of the power of combined spectral, temporal, and spatial informa-

tion can be illustrated by imagery from the Santa Monica Mountains region of southern

California. Ongoing research in this area is seeking improved ways to map vegetation patterns

and landscape processes, in part to improve models of fire behavior and assist resource

manager in this increasingly urbanized and disturbed region. Remote sensing is providing

maps of changing vegetation cover and land usage not attainable from ground surveys due to

the mountainous terrain, complex land ownership, and varying successional states and

community composition caused by frequent fire, agriculture, land development, and other

disturbances (Figure 22.3). NDVI images, derived from reflectance in two bands, the red and

NIR (Figure 22.4), readily depict relative levels of green vegetation cover across this land-

scape. However, a single overpass during spring, the peak biomass period for this landscape,

fails to clearly distinguish the several native vegetation types (Figure 22.5a), which for this

landscape include chaparral, coastal sage scrub, grassland, southern oak woodland, and

riparian woodland (Raven et al. 1986). In spring, these vegetation types have similar reflect-

ance spectra (Figure 22.5a) and are not readily distinguished by NDVI. However, these

vegetation types have contrasting seasonal patterns of green leaf display that can be readily

captured by multitemporal NDVI sampling (Gamon et al. 1995). This contrasting phenology

is readily revealed in the second image at the end of the summer drought (fall 1994),

illustrating a general decline in NDVI (darkening across the image) relative to the spring

image, with distinct patches emerging that reveal different vegetation types due to the

seasonal contrasts in canopy greenness (Figure 22.6), allowing NDVI to more fully separate

these vegetation types (compare Figure 22.5a and Figure 22.5b). In this way, the addition of

multitemporal sampling can reveal contrasting seasonal trajectories of canopy greenness that

are not evident in a single overpass.

In this landscape, key vegetation types can also be readily separated by more fully using

the rich spectral dimension present in hyperspectral imagery. With spectral mixture analysis,

which provides a tool for separating spectrally distinct landscape components, an individual

vegetation type (e.g., coastal sage scrub) can be readily depicted in an endmember frac-

tion map, in which varying intensity represents various quantities of that vegetation type

(Figure 22.5c). Elaborations of spectral mixture analysis that employ many spectral end-

members suggest that further separation of vegetation types into dominant species may be

possible (Roberts et al. 1998). Spectral mixture analysis is also able to depict relative levels of

green or dead biomass associated with vegetation composition, varying seasonal or succes-

sional states, or contrasting disturbance regimes (Gamon et al. 1993, Wessman et al. 1997,

Roberts et al. 1998). These improvements in the ability to distinguish vegetation types and

functional states are possible with a new generation of narrow-band imaging spectrometers

now available (Table 22.1). The ability to accurately depict dominant species or functional

vegetation types undoubtedly improves ecosystem models that require spatially explicit

vegetation maps as inputs.

At the global scale, multitemporal, broad-band satellite imagery capturing seasonal

variation in NDVI has often been used to develop improved global vegetation classifications

(DeFries and Townshend 1994a,b, DeFries et al. 1995, Sellers et al. 1996b, Nemani and

Running 1997). NDVI dynamics detected from satellite are providing valuable insights into

changing vegetation activity at global and decadal scales. For example, multiyear observa-

tions of AVHRR-derived NDVI for the continent of Africa have documented long-term

vegetation dynamics in the Sahel and have demonstrated cyclical patterns of Sahel vegetation

Francisco Pugnaire/Functional Plant Ecology 7488_C022 Final Proof page 665 18.4.2007 11:51am Compositor Name: VAmoudavally

Ecological Applications of Remote Sensing at Multiple Scales 665

(Tucker et al. 1991). This study is particularly notable because it counters the often-stated view

thatdesertification in this region is an irreversibleand inevitableprocess. However,newer findings

incorporating rain-useefficiencyinto this analysis suggest that the impacts of livestock grazing are

indeed causing long-term degradation of the region’s ecosystems (Hein and De Ridder 2006).

Long-term satellite records are also adding insights into biospheric responses to climate

change in northern latitudes. Recent satellite NDVI evidence suggests that early fingerprints

of global warming are now detectable by the increased vegetation activity in northern

latitudes (Myneni et al. 1997). One possible conclusion from these observations might be

that the northern regions are becoming more productive. However, most field studies indicate

FIGURE 22.5 Images of the Pt. Dume region of the Santa Monica Mountains of southern California.

(a) The NDVI in spring 1995; (b) the NDVI in fall 1994 (brighter areas indicate higher NDVI values,

signifying more green vegetation). Different native vegetation types, which include chaparral, coastal

sage scrub, annual grassland, and riparian areas, are difficult to separate in the spring (a), but are readily

distinguished in fall (b) due to the contrasting seasonal patterns of these vegetation types. Vegetation

types can also be readily separated with spectral mixture analysis, which models a landscape in terms

of endmember fractions, where each endmember represents a spectral type (vegetation type in this case).

with higher coastal sage scrub content). Images derived from NASA’s AVIRIS sensor (see Table 22.1).

In these images, north is to the right.

Francisco Pugnaire/Functional Plant Ecology 7488_C022 Final Proof page 666 18.4.2007 11:51am Compositor Name: VAmoudavally

666 Functional Plant Ecology

that the enhanced respiratory carbon loss from thawing northern ecosystems more than offset

any increase in vegetation productivity (Oechel et al. 1993, 2000). Although the exact

interpretation of these kinds of large-scale satellite data are subject to question due to the

difficulties of direct validation, they clearly indicate that remote sensing plays an increasingly

critical role in elucidating regional vegetation change in response to climate change and other

perturbations. Clearly, to fully interpret these changing patterns, improved validation at finer

scales are needed to supplement these emerging satellite tools. From this, we conclude that, to

address complex ecological questions, remote sensing is at its best when applied with ancillary

information obtained at a range of scales.

LINKING REMOTE SENSING TO PHOTOSYNTHETIC PRODUCTION

Satellite remote sensing now provides the frequent global coverage of surface reflectance

needed to drive models of global net primary productivity (Figure 22.7). The MODIS sensor

and the informatics system that makes the data freely available to the scientific community

(EOSDIS—NASA’s Earth Observing System Data and Information System) are revolutioni-

zing our ability to view changing patterns of global carbon exchange (Running et al. 2004).

However, as further discussed later, there are many issues of validation that still need to be

addressed. Multiscale remote sensing necessarily plays an important role in this validation.

In the past two decades, remote sensing has emerged as an essential tool for providing the

data fields that drive spatially explicit models of photosynthesis and net primary production

Wavelen

th (nm)

Riparian

Chaparral

Coastal sage scrub

600

0.0

0.1

0.2

0.3

0.4

Fall

Reflectance

Riparian

Chaparral

Coastal sage scrub

0.0

0.1

0.2

0.3

0.4

Spring

WBI

PRI

Reflectance

800 1000 1200

FIGURE 22.6 Spectra for representative types depicted in the AVIRIS image shown in Figure 22.5. Because

vegetation types are spectrally similar in spring, they are hard to distinguish from a single overpass during

that season. However, the contrasting spectral patterns emerging by fall allow ready separation of distinct

vegetation types. The relaxation of the slope in reflectance at the red edge (700 nm) indicates a drop in green

LAI as summer drought progresses. Note how the water absorption bands (near 970 and 1200 nm), which

are clearly visible in the spring, become less apparent in the fall spectra, indicating a progressive drying of the

landscape. Arrows indicate wavebands used for the PRI and the water band index (WBI). (Adapted from

Gamon, J.A., Lee, L.-F., Qiu, H.-L., Davis, S., Roberts, D.A., and Ustin, S.L., Summaries of the Seventh

Annual JPL Earth Science Workshop, Pasadena, California, 1998.)

Francisco Pugnaire/Functional Plant Ecology 7488_C022 Final Proof page 667 18.4.2007 11:51am Compositor Name: VAmoudavally

Ecological Applications of Remote Sensing at Multiple Scales 667

(NPP), and a wide variety of model formulations now exist. Most of these are variations of

the light-use efficiency model. In its basic form, the light-use efficiency model for terrestrial

vegetation states that the photosynthetic rate is a function of the absorption of photosyn-

thetically active radiation (APAR) and the efficiency with which that absorbed radiation gets

converted to fixed (organic) carbon:

Photosynthetic rate ¼ Efficiency APAR: (22:2)

In this case, the photosynthetic rate is the instantaneous photosynthetic rate. Although not

explicitly discussed here, this equation is most appropriately applied to the gross photosyn-

thetic rate, ignoring respiratory losses. To derive net photosynthesis, respiratory processes

(photorespiration and mitochondrial respiration) can be added as a separate term in the

model. Alternatively, the effects of respiration could be incorporated into the efficiency term.

If integrated over time (typically a growing season) and space (typically canopies, stands,

or regions), Equation 22.2 is often expressed as:

Primary productivity ¼ «  SAPAR, (22:3)

where primary productivity is usually expressed as NPP, often estimated by aboveground

biomass accumulated in a growing season, SAPAR is the annual integral of radiation

absorbed by vegetation, and « represents the efficiency with which absorbed radiation is

converted to biomass (Monteith 1977). Again, the respiratory contribution of nonphotosyn-

thetic organs (e.g., stems, branches, and roots) can be incorporated into «, or more explicitly

June 2002

December 2002

0123

Net primary productivity (kgC m

−2

year

−1

)

FIGURE 22.7 (See color insert following page 684.) Global NPP estimates for the months of June and

December, 2002, derived from the MODIS satellite sensor. (From NASA’s Earth Observatory,

http:== earthobservatory.nasa.gov=)

Francisco Pugnaire/Functional Plant Ecology 7488_C022 Final Proof page 668 18.4.2007 11:51am Compositor Name: VAmoudavally

668 Functional Plant Ecology

included as a separate respiratory term or coefficient (Prince and Goward 1995, Ruimy et al.

1996, Landsberg et al. 1997).

Equation 22.2 and Equation 22.3 can be viewed as a simple conceptual model, or can be

further elaborated as a more explicit, mechanistic model, in which case, the addition of

distinct terms for autotrophic (and possibly heterotrophic) respiration would be appropriate.

Similarly, simplistic versions of this model might represent the efficiency («) term as a fixed

coefficient, whereas more mechanistic versions of this model formulation might treat effi-

ciency as a variable that is continually affected by environmental factors. The APAR and

maximum efficiency («) values determine the maximum attainable photosynthetic rate, some-

times called the potential photosynthetic rate. Environmental or physiological factors that

reduce photosynthetic rate through a variety of mechanisms impact either APAR or light-use

efficiency, or both. For example, water stress, temperature extremes, and nutrient limitations

can all reduce effective LAI and thus absorbed radiation (APAR). These same conditions can

reduce light-use efficiency through downregulation of photosynthetic processes involving

stomatal closure, enzyme inactivation, and altered light energy distribution involving photo-

protective mechanisms (Bjo

rkman and Demmig-Adams 1994, Gamon et al. 1997, 2001;

Figure 22.8).

The beauty of the light-use efficiency model is that it can be readily applied at several

temporal and spatial scales and can be readily linked to remote sensing, which is also scale-

able, allowing application and validation at multiple spatial and temporal levels. The remote

sensing link is usually provided by the APAR term, which can be further defined as the

product of irradiance of photosynthetically active radiation (PAR) and the fraction of

the irradiance that is absorbed by photosynthetic (i.e., green) canopy elements (F

APAR

APAR ¼ F

APAR

 PAR: (22:4)

Both PAR irradiance (Frouin and Pinker 1995) and F

APAR

can be determined from remote

sensing. F

APAR

can be derived from NDVI or other vegetation indices, and this relationship

has been well tested, both with theoretical studies (Kumar and Monteith 1981, Asrar

et al. 1984, Sellers 1987, Prince 1991) and empirical measurements at many spatial and

PAR

Chl

Heat

(xanthophyll cycle)

Fluorescence

Electron transport

Carboxylation

Photorespiration

FIGURE 22.8 Schematic depicting possible fates of PAR absorbed by photosynthetic pigments (Chl).

Absorbed energy can be used to drive photosynthetic electron transport and carbon uptake via

carboxylation. Under conditions of reduced photosynthetic light-use efficiency, the photosynthetic

system downregulates, and an increased proportion of absorbed energy is dissipated by fluorescence

or heat production. The operation of the xanthophyll cycle is linked to this heat dissipation (Pfu

ndel and

Bilger 1994, Demmig-Adams and Adams 1996, Demmig-Adams et al. 1996). Both xanthopyll pigment

conversion and chlorophyll fluorescence provide useful means for optically detecting reductions in

photosynthetic efficiency. The relative levels of photosynthetic pigments (chlorophylls and carotenoids)

provide additional indicators of photosynthetic activity.

Francisco Pugnaire/Functional Plant Ecology 7488_C022 Final Proof page 669 18.4.2007 11:51am Compositor Name: VAmoudavally

Ecological Applications of Remote Sensing at Multiple Scales 669