Principles of radiation interception and use

26 C. Black, C. Ong Agricultural and Forest Meteorology 104 2000 25–47 by showing that accumulated dry matter production of a wide range of crops and orchards in Britain was linearly related to accumulated intercepted solar radi- ation Monteith, 1977a. The second was the applica- tion of the thermal time concept to describe the effects of temperature on crop development Monteith, 1979. These concepts were tested in the tropics in close collaboration with the International Crop Research Institute for the Semi-Arid Tropics ICRISAT, Hy- derabad, India, where JLM was Director of the Re- source Management Programme between 1987 and 1991, to assist in the more practical problem of es- tablishing genotype–environment interactions for mil- let and groundnut Williams, 2000 and determine the mechanisms responsible for overyielding in intercrop- ping systems Marshall and Willey, 1983; see also re- views by Fukai and Trenbath, 1993; Ong and Black, 1994; Azam-Ali, 1995. Today these concepts have been applied throughout the tropics and subtropics and such studies are no longer confined to international centres; for example, Vijaya Kumar et al. 1996 ap- plied these concepts to castor beans at the Central Re- search Institute for Dryland Agriculture, Hyderabad, India. JLM’s resource capture concept has also been applied in forestry research Landsberg, 1986; Can- nell et al., 1987, crop simulation models Ritchie and Otter, 1985; Jones and Kiniry, 1986, as a tool for biomass prediction in agronomic research Sinclair et al., 1992 and in remote sensing Christensen and Goudriaan, 1993. Recently, the same approaches have been used to unravel the more complex interactions between trees and crops in agroforestry systems in studies of both above- and below-ground interactions Ong et al., 1991; Ong and Black, 1994. Initial findings indicated that there are critical differences between intercrop- ping and agroforestry which are apparently linked to the relative importance of below-ground interactions Ong et al., 1996. Even more formidable difficulties were encountered when these essentially agronomic plot size findings were extrapolated to farm and land- scape levels since the extensive lateral growth of tree roots may lead to their extension across farm bound- aries or between adjacent experimental plots Hauser, 1993; Rau et al., 1993. In this review, we describe how John Monteith’s concepts of resource capture and thermal time have been applied in tropical agricultural research and focus on recent attempts to apply these concepts to agroforestry. Finally, we highlight some of the progress made and the new challenges ahead.

2. Principles of radiation interception and use

The capture of radiation and its use in dry mat- ter production depends on the fraction of the incident photosynthetically active radiation PAR that is in- tercepted and the efficiency with which it is used for dry matter production. Intercepted radiation S i is of- ten estimated as the difference between the quantity of incident radiation S and that transmitted through the canopy to the soil S t . However, this approach has inherent technical and theoretical difficulties since is does not account for the reflection of incident radiation from the canopy surface typically 5–20 depending on surface characteristics and moisture content, or for radiation intercepted by non-photosynthetic canopy el- ements. As a result, interception by photosynthetically competent tissues may be greatly overestimated, par- ticularly for canopies which are senescing or contain numerous woody structural elements. Corrections for these errors have often been ignored when estimating S i and photosynthetic efficiency. The quantity of radiation intercepted depends on the amount received by the canopy, its size and dura- tion and fractional interception f . The seasonal time course of f, defined here as S i S, varies greatly de- pending on canopy architecture and the phenology of the vegetation involved; thus, f increases more rapidly in cereals such as sorghum Sorghum bicolor than in legumes such as groundnut Arachis hypgaea, reflect- ing their differing rates of leaf initiation and expan- sion Squire, 1990. The variation in f between crops is generally smaller than that in green leaf area index, partly because the extinction coefficient for radiation k is often larger in species whose canopy expands slowly; maximum f values may therefore differ little between crops grown under non-limiting conditions. Mean f values calculated over the duration of the crop ¯ f are generally lower in short-duration cereals ca. 0.5 and legumes ca. 0.15 than in perennial species ca. 0.9, largely because of the differing duration of ground-cover Squire, 1990. The theory and experimental approaches required to quantify radiation interception by monocultures C. Black, C. Ong Agricultural and Forest Meteorology 104 2000 25–47 27 containing uniform plants are now well-established and have been widely used in both temperate and tropical environments during the past three decades Monteith et al., 1994. These approaches assume that the canopy is a heterogeneous arrangement of ran- domly distributed leaves, and that the spatial variation in radiation transmission by the canopy is limited. However, these conditions are clearly not met in inter- crops or agroforestry systems because of the extensive horizontal and vertical variation in canopy structure introduced by the intimate mixture of species with differing planting dates and arrangement, heights and maturity dates. Canopy architecture is also constantly changing in mixed cropping systems because of the differing growth rates and canopy durations of the component species. For example, compact legumes growing adjacent to taller cereals initially experience greater competition than in the equivalent monocrop because of the faster growth of the cereal, but subse- quently experience less competition for much of the reproductive phase due to the earlier harvest of the cereal component. The methodological problems involved in charac- terising the spatial and temporal variation in radiation interception are much greater in mixed communities than in monocultures, and the partitioning of radia- tion interception and also water uptake between the components of such systems has provided a major challenge. Detailed reviews of radiation interception in intercropping and agroforestry systems are given by Keating and Carberry 1993 and Ong et al. 1996. This spatial variability is relatively modest and ca- pable of resolution in annual row intercrops, as has been demonstrated by several studies e.g. Marshall and Willey, 1983; Azam-Ali et al., 1990, but is much greater in agroforestry systems, particularly those in- corporating large overstorey trees. An additional com- plexity is the extended time scale and ever-changing relationship between the tree and crop components. In annual intercrops, the growing season is normally confined to a period of 100–150 days, whereas the trees in agroforestry systems are often active through- out the year. The size of the trees and their influence on associated crops may also change rapidly from year to year. For example, in a recent trial at ICRAFs Machakos Research Station in Kenya, Grevillea robusta trees reached a height of 6–8 m within 5 years of planting. In 1981, John Monteith Monteith, 1981 suggested that the basic principle underlying resource capture in mixed communities is that the complementarity or competitiveness of interactions between species will depend on their ability to capture and use the most essential limiting growth resources effectively. This theory has been the cornerstone of numerous stud- ies of resource partitioning between the component species of intercrops and agroforestry systems by his team at ICRISAT and other groups. Thus, the capture of the limiting resource i.e. light, water or nutrients will depend on the number, surface area, distribution and effectiveness of the individual elements of the canopy or root system of the monocrop or mixture being examined. One of the earliest detailed studies of resource par- titioning in agroforestry systems was that described by Monteith et al. 1991 and Corlett et al., 1992a,b. In this study in India involving a Leucaena leuco- cephalamillet Pennisetum americanum alley crop- ping system, tube solarimeters of a design originally developed under JLM’s guidance Green and Deuchar, 1985 were used to determine radiation interception by both species Table 1. Total intercepted radia- tion during the 1986 rainy season was substantially greater in the alley cropping system than in either of the monocrops, primarily because the presence of leu- caena increased fractional interception during the early stages of the growing season, while the millet provided a more complete ground cover across the alleys during the later stages of the season. Biomass production was also substantially greater in this treatment. The alley millet intercepted 48 less radiation than monocrop millet due to its lower population on a system area ba- sis and the shading effect of the leucaena. However, the reduction in biomass production by alley-cropped millet was smaller than that in radiation interception due to an increase in conversion efficiency e, possi- bly because the light saturation of photosynthesis as- sociated with drought occurred less frequently under partial shade. The conversion coefficient, defined here as the quantity of biomass produced per unit of inter- cepted radiation g MJ − 1 , provides a measure of the ‘efficiency’ with which the captured radiation is used to produce new plant material; the alternative term ra- diation use efficiency RUE is also commonly used. In 1986, the hedges were regularly pruned to a height of 70 cm to minimise shading of the millet, 28 C. Black, C. Ong Agricultural and Forest Meteorology 104 2000 25–47 Table 1 Intercepted total solar radiation, above-ground biomass produc- tion and biomass production per unit of intercepted radiation e modified from Corlett et al., 1992b Intercepted radiation MJ m − 2 Biomass t ha − 1 e g MJ − 1 Rainy season July– August 1986 Sole millet 581 4.7 0.81 Alley millet 300 3.1 1.03 Sole L. leucocephala 520 4.0 0.77 Alley L. leucocephala 510 4.0 0.77 Total alley system 810 7.1 0.81 Dry season September 1986–June 1987 Sole L. leucocephala 1270 1.5 0.12 Alley L. leucocephala 1160 1.7 0.15 Year July 1986– August 1987 Total alley system 1970 8.8 0.45 Rainy season July– August 1987 Sole millet 504 5.0 0.98 Alley millet 180 0.9 0.50 Sole L. leucocephala 861 7.1 0.82 Alley L. leucocephala 748 6.4 0.86 Total alley system 928 7.3 0.79 but in 1987 they were allowed to grow unchecked, as is reflected by the substantially higher radiation in- terception by leucaena in all treatments and the 64 reduction in interception by the alley-cropped millet relative to monocrop millet. The intense shading of the former was accompanied by a sharp reduction in e and an 82 reduction in above-ground biomass. These results reinforce the conclusion reached in other studies that, although leucaena may be a successful component of agroforestry systems in the humid trop- ics, it is too competitive for successful adoption in water-limited semi-arid regions Singh et al., 1989; Monteith et al., 1991. When growth is not limited by water or nutrient sup- plies, the quantity of biomass produced by monocrops is limited primarily by the quantity of radiation cap- tured. JLM suggested that seasonal biomass accumu- lation for a given species may be expressed as the time integral of the product Sf i e, where S denotes inci- dent radiation, f i the fractional interception on a given day, and e the conversion coefficient for radiation as defined above Monteith et al., 1991. These authors suggested that this philosophy was applicable both to monocrops and to the components of intercropping and agroforestry systems, provided f i and e could be determined for each component. They also proposed that, if fractional interception by the whole system was recorded, an overall system average value for e could be obtained and system performance analysed in terms of the efficiency of radiation capture and utilisation to produce biomass. The use of the acquired resources was therefore assumed to depend on the conversion coefficient of the species involved and environmental influences such as drought. A major advantage of ex- pressing productivity in these terms is to emphasise the apparent conservativeness of e under many con- ditions. This approach has subsequently been widely adopted in studies of resource partitioning in inter- cropping and agroforestry. Numerous studies of annual crops, and some with perennial species, have demonstrated the existence of close correlations between dry matter production and cumulative intercepted radiation. For example, Stirling et al. 1990 examined the impact of artifi- cial shade imposed on groundnut Arachis hypogaea between the onset of peg initiation or pod-filling and final harvest using bamboo screens. A close linear correlation between above-ground biomass and cu- mulative intercepted radiation was found in all treat- ments, although the quantity of biomass produced per unit of intercepted radiation was substantially greater when shading was imposed from peg initia- tion onwards Fig. 1. In the absence of stress, e is often conservative, typically ranging between 1.0 and 1.5 g MJ − 1 for C3 species in temperate environments Monteith and Elston, 1983; Russell et al., 1988, 1.5–1.7 g MJ − 1 for tropical C3 species Kiniry et al., 1989; Monteith, 1990 and up to 2.5 g MJ − 1 for trop- ical C4 cereals under favourable conditions Squire, 1990. However, the work of Stirling et al. 1990 showed that e may vary substantially within a single season between 0.98 g MJ − 1 in the unshaded control and 2.36 g MJ − 1 in crops shaded from peg initiation onwards. Thus, plants in the latter treatment inter- cepted approximately one-quarter of the radiation received by the unshaded control, but converted this to dry matter 2.4 times more efficiently. In contrast, in a study of millet grown in a Leucaena leucocephala alley-cropping system in India, Corlett C. Black, C. Ong Agricultural and Forest Meteorology 104 2000 25–47 29 Fig. 1. Relation between dry matter production and cumulative intercepted radiation between 44 and 98 days after sowing for groundnut grown at Hyderabad, India. 3 unshaded control; 1 and 2 plants subjected to 50 artificial shade between peg initiation or the onset of pod filling and final harvest modified from Stirling et al., 1990. et al. 1992b observed that a sharp inflection in the relationship between biomass accumulation and radia- tion interception occurred around anthesis, after which the slope of the relationship decreased sharply and lit- tle further biomass was produced in either monocrop or alley-cropped millet in either of the years examined Fig. 2. These results demonstrate that the relation- ship between biomass accumulation and intercepted radiation and the derived values for e may vary de- pending on the time scale over which the measure- ments are made, particularly when periods near the end of the cropping cycle are included, when radia- tion is increasingly being intercepted by senescent or dead tissues; indeed, total dry matter may decrease as crops approach maturity and leaves and other organs are lost. Radiation interception by non-photosynthetic tissues may be even greater in trees, where branches and other structural elements may account for a sig- nificant proportion of the intercepted radiation. Corlett et al. 1992b concluded that over half of the reduction in the pre-anthesis conversion coefficient observed in alley-cropped millet was attributable to the reduced PAR content of the radiation reaching the shaded crop. Subsequent resource capture studies at ICRISAT examined the partitioning of water and light in peren- Fig. 2. Relation between accumulated above-ground dry matter and intercepted radiation for a sole millet and b millet alley-cropped with Leucaena leucocephala at Hyderabad, India. Radiation con- version coefficients calculated for the pre- and post-anthesis periods were, respectively, 1.80 r 2 = 0.99 and 0.13 r 2 0.50 g MJ − 1 for sole millet in 1986, 1.46 r 2 = 0.99 and 0.53 r 2 = 0.96 g MJ − 1 for sole millet in 1987, 1.72 r 2 = 0.96 and −0.04 r 2 0.50 g MJ − 1 for alley-cropped millet in 1986, and 1.09 r 2 0.96 and 0.33 r 2 0.50 g MJ − 1 for alley-cropped millet in 1987 modified from Corlett et al., 1992b. nial pigeonpeagroundnut Cajanus cajunArachis hypogaea agroforestry systems Marshall, 1995; Ong et al., 1996. Monocrops of each species were compared with line-planted and dispersed mixtures containing identical pigeonpea populations. At its initial optimum density of 9 plants m − 2 in 1989, the 1-year-old sole pigeonpea captured over seven times as much radiation 1960 MJ m − 2 as the line-planted and dispersed pigeonpea, with their lower initial pop- ulation of 0.5 plants m − 2 . Even when interception by groundnut was included, interception by monocrop pigeonpea was approximately double the totals for 30 C. Black, C. Ong Agricultural and Forest Meteorology 104 2000 25–47 the line-planted and dispersed systems. However, the seasonal mean e value for monocrop pigeonpea was lower than that for line-planted or dispersed pigeon- pea 0.42, 0.54 and 0.56 g MJ − 1 , respectively, with the result that its substantial interception advantage was not reflected by a similar biomass advantage. Monocrop groundnut intercepted 10–40 more radia- tion than line-planted or dispersed groundnut in 1989, although this was again not matched by a similar increase in biomass. The population of the monocrop pigeonpea was re- duced to 0.5 plants m − 2 in 1990 because the initial population was believed to exceed the optimum for maximum productivity during the second year, when competition between neighbouring trees was expected to cause extensive mortality; this reduction also estab- lished identical populations in all treatments. Intercep- tion by the 2-year-old monocrop pigeonpea was 70 greater than that for line-planted pigeonpea, but al- most 50 lower than that for the dispersed pigeonpea, a pattern reflected by biomass production Ong et al., 1996. The combination of greater radiation capture and lower e in the dispersed pigeonpea probably re- sulted from its bushy structure, which would have in- creased the fraction of radiation intercepted by stems and branches. The much greater total interception by the dispersed system in 1990 was primarily due to the pigeonpea which captured over 80 of the incident ra- diation, although its substantial interception advantage was not reflected by increased biomass production be- cause of its relatively low e value. These observations are analogous to the leucaenamillet system described by Corlett et al. 1992b, in which the less efficient tree component captured most of the incident radiation dur- ing the second year, to the detriment of crop growth. The observed variability in experimentally deter- mined e values contrasts with earlier views that e is highly conservative except during severe water stress Azam-Ali et al., 1989, but complies with more re- cent suggestions that the assumption of a constant value within species or cultivars may be erroneous Demetriades-Shah et al., 1992, 1994. Indeed, these papers advanced some controversial views which sparked considerable lively debate. These authors comprehensively criticised the concept that biomass accumulation may be linked directly with cumu- lative intercepted radiation, and that meaningful e values may be derived from such correlations. They argued that the concept of radiation use efficiency is over-simplistic, cannot improve our understanding of crop growth and is of limited value in predicting yield, and concluded that there was little evidence that incident radiation is a critical limiting factor de- termining crop growth under normal field conditions. In support of this argument, Demetriades-Shah et al. 1992 presented data to show that the growth of broiler chickens could be correlated with estimates of radiation interception derived from the size and popu- lation of birds, even though the two variables were not causally related. They also contested the validity of establishing correlations between cumulative values for biomass and radiation on the grounds that plot- ting cumulative values for randomly selected pairs of numbers could produce apparently significant corre- lations, even though there was no linkage between the numbers involved. They re-evaluated several experi- mental datasets and concluded that analysis of crop growth in terms of cumulative intercepted radiation and the conversion efficiency of solar energy during dry matter production should be approached with caution. A major plank in their argument was that photosynthesis, and hence crop growth rate, depends on numerous soil, atmospheric and biological factors of which radiation is only one component. They sug- gested that good correlations will always be found be- tween radiation interception and any growing object, even when radiation is not the limiting variable, and so a close correlation between crop growth and radi- ation interception may be expected even when light is not a major limiting factor. Thus, although solar energy may be the most fundamental natural resource for crop growth from a physical viewpoint, from a bi- ological viewpoint it is no more important than water, nutrients, CO 2 or any other essential commodity. As a result, analysis of crop growth in terms of its radiation conversion coefficient may be inappropriate when variables other than radiation are the primary limiting factor. Vijaya Kumar et al. 1996 provided experimental support for this view when they showed that the con- version coefficient for rainfed castor beans Ricinus communis was less stable than previously suggested. The values obtained varied from year to year and were influenced by sowing date, decreasing with lateness of planting within the range 0.79–1.10 g MJ − 1 ; val- ues recorded prior to flowering were more stable than C. Black, C. Ong Agricultural and Forest Meteorology 104 2000 25–47 31 those obtained after flowering began. These variations were positively correlated with atmospheric satura- tion deficit D and wind speed and negatively corre- lated with water availability and temperature. In view of these results, Vijaya Kumar et al. suggested that it would be necessary to incorporate the influence of weather conditions into the algorithms used to calcu- late biomass in crop simulation models in order to pre- dict effects on growth and yield. Other authors have previously described the influence of physical vari- ables such as atmospheric saturation deficit Stockle and Kiniry, 1990, temperature Hammer and Vander- lip, 1989, water stress Ong and Monteith, 1985 and biological factors such as phenology Spitters, 1990; Giauffret et al., 1991, CO 2 fixation pathway and fol- iar nitrogen content Sinclair and Horie, 1989 on e. In a spirited defence of the validity, generality and robustness of correlations between intercepted radia- tion and growth and the conservativeness of e, JLM concluded that few of the arguments advanced by Demetriades-Shah et al. 1992 were invincible, but that their paper would provoke crop scientists to con- sider more carefully how errors involved in measur- ing intercepted radiation may be minimised and how responses to stress may be interpreted in terms of sea- sonal changes in f and e. JLM stressed that, while some early papers concerned with the relationship between radiation capture and dry matter production, including his own 1977a contribution, suggested that e was con- sistent between species within major groups and was apparently less affected by stress than f, many subse- quent reports showed that e is not invariably conserva- tive, but may vary depending on growth stage, water and nutrient availability and the incidence of disease Green, 1987; Garcia et al., 1988; Steinmetz et al., 1990; Bastiaans and Kropff, 1993. JLM suggested that, contrary to the view of Demetriades-Shah et al. 1992, this did not invalidate the concept but instead highlighted the need to test and improve methodology as new information becomes available. In their original paper, Demetriades-Shah et al. suggested that relative growth rate RGR or net assimilation rate NAR may be preferable to us- ing cumulative radiation interception and conversion coefficients when analysing crop growth. However, Monteith 1994 argued that, since RGR and NAR both decline systematically during crop growth, any attempt to correlate these variables with climatic or environmental factors is likely to meet with lim- ited success. A reanalysis of data first published by Kanemasu et al. 1990 was used to support this argument. JLM concluded that there is a sound mechanistic basis for observations that biomass ac- cumulation often bears an almost linear relationship to intercepted radiation since much of the radiation captured by annual crops is intercepted by young leaves with a photosynthetic efficiency that is almost constant over a wide range of irradiances, stressed or senescent leaves spend much of their time on the linear phase of the photosynthetic light response curve, and there is often little seasonal variation in the average point at which leaves operate on the light response curve. In his original discussion of the concept conversion coefficients, JLM 1977a presented the results of a theoretical analysis which demonstrated that e would be expected to be directly related to the CO 2 exchange rates of leaves, while Sinclair and Horie 1989 subsequently described the steps involved in the theoretical derivation of daily e values. Their analysis demonstrated the close link- age between e and the light-saturated rate of CO 2 exchange, the CO 2 fixation pathway and respiratory characteristics. JLM’s arguments were supported by Kiniry 1994 and Arkebauer et al. 1994, who suggested that Demetriades-Shah et al. 1992 had overlooked the fact that many environmental stresses which limit growth act through physiological pathways directly involving the photosynthetic process and its products. Arkebauer et al. 1994 advanced strong arguments to show that e cannot be expected to be constant, even within a single species or genotype, in the face of changes in other environmental variables. They sug- gested that ‘the concept of RUE is therefore a very powerful tool in understanding crop growth and pre- dicting crop yield’ since factors which affect canopy and leaf photosynthesis and crop biomass accumula- tion have a well defined influence on the value of e expressed by the crop. Arkebauer et al. 1994 also dismissed criticisms that the reported variability in e is symptomatic of inherent inadequacies in the concept by arguing that these resulted largely from an inconsistent definition of e and lack of consideration of its physiological basis. The definition of e involves three separate fac- tors: firstly, the type and energy content of the carbon 32 C. Black, C. Ong Agricultural and Forest Meteorology 104 2000 25–47 involved, i.e. net CO 2 uptake by the canopy, total above-ground dry matter production, or total plant dry matter including roots and storage organs; sec- ondly, the way in which radiation is characterised i.e. total incident solar radiation, intercepted shortwave radiation, intercepted PAR or absorbed PAR; finally, the time scale over which e is calculated is extremely important and may range from instantaneous through hourly, daily or weekly estimates to the seasonal time scale. Because widely differing definitions of e have been adopted, the values obtained may be expected to show substantial variation, contributing to the contro- versy over its conservatism. Thus, Begue et al. 1991 found that, when e was calculated for monocrop mil- let over 5 day intervals, the values declined during the season from ca. 5.6 to 0.5 g MJ − 1 PAR when expressed on a total above-ground basis, but the vari- ation was reduced when e was calculated for specific developmental stages, ranging from 5.0 g MJ − 1 dur- ing tillering to 2.3 g MJ − 1 during maturation. Ong and Monteith 1985 also summarised values calcu- lated over extended periods and showed that e was highly conservative for three well watered pearl millet crops, ranging from 2.15 to 2.37 g MJ − 1 during the pre-anthesis period; the values for two water-limited crops were 2.0 and 1.5 g MJ − 1 . The seasonal values were lower, ranging from 1.26 to 1.49 g MJ − 1 for well watered crops and 1.14–1.17 g MJ − 1 for water-limited stands. Adverse environmental conditions may therefore reduce e because of their adverse effect on photo- synthetic activity. Although Demetriades-Shah et al. 1992 pointed out that stress is commonly encoun- tered in the field, Arkebauer et al. 1994 empha- sised that, far from invalidating the approach, a major strength of e is that it can be used to quantify the im- pact of stress factors by comparing the observed values with those obtained under unstressed conditions. They also suggested that, if radiation interception can be es- timated from models of canopy development, the cu- mulative product of intercepted radiation and e calcu- lated on a daily basis would enable maximum growth and yield potentials to be estimated for specific crops and environments. This approach has proved useful in assessing the yield potential of various annual crops including maize Zea mays; Jones and Kiniry, 1986, soybean Glycine max; Spaeth et al., 1987 and wheat Triticum aestivum; Amir and Sinclair, 1991a. Modelling crop yields under water stress condi- tions is important in understanding their growth un- der normal field conditions. As indicated previously, e is influenced by stress factors because of its depen- dence on the gas exchange characteristics of leaves, which are in turn related to soil moisture status. Simple models in which the impact of soil water deficits has been related directly to e, crop growth and yield have been used successfully to simulate growth in soybean Muchow and Sinclair, 1991, maize Muchow and Sinclair, 1991 and wheat Amir and Sinclair, 1991b. For instance, Sinclair et al. 1992 used a soybean model incorporating e as a function of soil water con- tent to compare yields across years and locations with differing rainfall in Argentina. Remotely-sensed, as opposed to ground-level mea- surements, of annual PAR interception or even simple accumulated values for spectral vegetation indices have been correlated with annual biomass production for a range of ecosystems including crops Daughtry et al., 1992 and semi-arid grasslands Prince and Tucker, 1986, and similar approaches have been applied at the continental Goward et al., 1985 or global scale Potter et al., 1993; Ruimy et al., 1994. This approach was adapted by Goetz and Prince 1996 who used a simplified canopy radiative transfer model in combination with high resolution LAND- SAT greeness vegetation index images to estimate annual interception of PAR by stands of quaking aspen Populus tremuloides and black spruce Picea mariana in Northeastern Minnesota. The relation- ship between estimated annual PAR interception and measured above-ground biomass production was then used to calculate e. Annual PAR interception was generally higher and more consistent between stands of aspen 600–1100 MJ m − 2 per year than in black spruce 100–1100 MJ m − 2 per year, for which the values varied widely irrespective of age. Stand age influenced the relationship between cumulative PAR interception and biomass production in aspen, with a strong correlation being found in young stands and a weak correlation in mature stands, an effect attributed to increasing maintenance respiration de- mands as the ratio of foliar to total biomass decreased with age. The e values for aspen 0.44–1.29 g MJ − 1 with a mean of 0.92 g MJ − 1 were lower than those for black spruce 0.17–0.89 g MJ − 1 with a mean of 0.49 g MJ − 1 . C. Black, C. Ong Agricultural and Forest Meteorology 104 2000 25–47 33

3. Principles of water uptake and use