Thinking about Efficiency of Resource Use in Forests

The growth of forests can be described as a function of the supply of resources, the proportion of these resources captured by trees, and the efficiency with which trees use these resources to fix carbon dioxide. This function can be modified to explain wood production by subtracting the allocation of carbon (C) to respiration and other tissues. At the scale of leaves and seconds, rates of net photosynthesis typically show declining marginal gains in carbon (C) with increasing rates of light absorption, water transpiration, and sometimes nitrogen (N) concentration. However, these trends may not represent the trends that occur at the scale of forests and years, owing to more complete C accounting (including costs of synthesis and maintenance of tissues), and interactions among resources. Patterns in the growth of forests, across environmental gradients or silvicultural treatments, demonstrate substantial variation in efficiency of resource use at the scale of forests and years, including increasing efficiency of resource use as rates of resource use increase. Case studies from Eucalyptus plantations indicate that more productive sites tend to have higher efficiency of resource use than less productive sites, and within-site increases in production from silvicultural treatments may result in part from increased efficiencies in resource use. The questions raised here apply to all forests, but the level of confidence in our general conclusions remains limited by the number of studies available with complete estimates of rates of resource use and production. Introduction Forest production depends on trees obtaining resources from the environment and using these resources to fix atmospheric CO2 into biomass. The production of wood also depends on the pattern of biomass allocation in trees; wood biomass commonly accounts for 10 to 30% of the total production of trees. This verbal model can be stated in a quantitative form that provides some powerful insights into patterns in forest growth across species, environmental gradients, and stand age: Gross Primary Production (GPP) = Resource Supply x Proportion of Resource Supply Captured x Efficiency of Resource Use (based on Montieth 1977). This equation can be modified to define the production of woody biomass as the same function, minus allocation to other tissues and respiration. Several authors have used versions of this equation to explain patterns in forest growth (cf. Cannell 1989, Binkley et al. 1990, Landsberg 1997, Ryan et al. 1997, McMurtrie et al. 1997), but wider use of this approach would be useful. One forest may produce more wood than another as a result of higher resource supply, by capturing a greater proportion of available resources, by using resources more efficiently, or by allocating a greater proportion of biomass to wood. Foresters commonly expect higher rates of wood production from sites with greater supplies of resources such as water and nutrients, and forest scientists have quantified the rates of resource use for many forests around the world. Expectations about patterns of resource use efficiency have been clouded by incomplete production budgets, confusion over scales and definitions of terms, and poor definitions of economic analogies. In this paper, we briefly highlight the utility of the production ecology equation, and focus on ways of thinking about the efficiency of resource use. The production ecology equation An application of the production ecology equation (Figure 1) details the components of the increase in production from irrigation of a clonal Eucalyptus stand in Brazil. Irrigation increased GPP from 6.2 kg m yr to 11.3 kg m yr. Irrigation did not alter the supply of incoming light, but the percentage of light intercepted by the canopy increased from 63% to 71%. This increase in light capture is notably smaller than the increase in GPP, indicating a substantial increase in the efficiency of converting captured light into biomass. Irrigation increased the annual supply of water from 1.21 m/m to 2.17 m/m, but the percentage of the water supply actually used by the trees declined from 74% to 58%, giving an actual annual water use of 0.90 m/m for the rainfed stand and 1.25 m/m for the irrigated stand. The difference in water use was again smaller than the relative increase in GPP, indicating a substantial increase in the efficiency of water use by the irrigated stand (rising from 6.9 kg/m to 9.0 kg/m). Irrigation more than doubled wood production from 1.44 kg m yr to 3.46 kg m yr, a larger proportional increase than the increase in GPP (Figure 1), and a greater increase in the efficiency of wood production per unit of resource used. Patterns of Resource Use Efficiency The example above showed that the efficiency of using light and water increased as the amount of light and water used by the forest increased. This may seem counter-intuitive; many ecologists expect “declining marginal returns” of carbon gain per unit of resource used as the supply of a resource increases. This may be a logical expectation for some situations, but this pattern may not describe forest resource use at annual time scales. We briefly discuss prior work on resource use efficiency, and then explore present specific case studies. Discussions of efficiency are often clouded by differences in terminology; Table 1 provides a definition of terms commonly used in discussions of forest production and resource use. The expectation of declining marginal returns is widespread in biology and chemistry, such as the Michaelis-Menton equation for the kinetics of enzyme reactions. Indeed, Pastor and Bridgam’s (1999) discussion of resource use efficiency uses the word “law” when referring to expectations of declining marginal returns. As we elaborate below, this expectation may describe resource use efficiency very well at the scale of leaves and seconds (where reactions are most analogous with simple chemical reactions), but may not encompass the ecosystemscale processes that determine efficiency of resource use at the scale of forests and years. This discrepancy across scales may derive from a broader accounting of the production budget (including costs of generating leaves and roots), and from interactions among the supplies and efficiency of use for several resources. This classic expectation can be illustrated at the scale of leaves at the time scale of one second (Figure 2). At very low rates of resource use, leaves do not show positive rates of net photosynthesis; this is the classic “compensation point” from plant physiology, or Rmin in the terminology of Pastor and Bridgham (1999, which provides thorough coverage of the theoretical implications of resource efficiencies). When the resource use level is above this minimum, net photosynthesis increases rapidly with increases in resource use, the rate of increase typically slows as resource use rates become very high. The slope of the curve of resource use efficiency with increasing resource use is steep when resource use is close to the Rmin, and then declines as resource use moves far beyond the Rmin. These general expectations are illustrated at the scale of leaves and seconds for 5-year-old Eucalyptus saligna in a plantation in Hawaii (Figure 2). This general pattern indicates that the efficiency of using water and light (defined in units of production per unit of resource used) must initially increase as the rate of resource use increases, then decline. The efficiency of N use (at the scale of seconds) differs conceptually from that of light and water, as the N acts as a catalyst (not consumed in the reaction of photosynthesis) rather than as a resource; in this case, the light-saturated rate of photosynthesis increased linearly with increasing N concentration in leaves, with a Yintercept of 0, indicating a constant rate of N use efficiency across levels of N concentration. Moving from the scale of a leaf to the scale of a forest and a year, the key issues in resource-use efficiency center on two conditions: 1) whether the X intercept is positive (= negative Y intercept, or a positive Rmin), in the relationship between production and resource use, and 2) whether resource use efficiency increases with the rate of resource use rises to the point where efficiency declines (as illustrated at the scale of leaves and seconds in Fig. 2). We expect condition #1 to be universal, because forests are not observed to occur in environments with extremely low rates of resource supply (and hence use), and because the production and maintenance of treesized tissues represents a very large “investment” that must be covered by obtaining a large minimum supply of resources. The instantaneous light compensation point for a leaf is always much lower than the “compensation point” that would provide for the C invested in constructing the leaf. Net growth of trees would occur only after construction and maintenance costs have been covered. We think condition #2 remains remarkably underinvestigated for forests. Too few estimates have been developed for complete C budgets and resource use at the scale of hectares and years. In the absence of evidence at this scale, ecologists and foresters often assume strong declines in the marginal gain of C for marginal increases in resource use, much like the pattern that has been observed routinely at the scale of leaves and seconds. The limited information we have examined suggests that in fact forests often remain in the range of production and resource use where efficiency of resource use continues to increases. As noted by Pastor and Brigham (1999), a positive Xintercept must give a positive slope for the relationship between growth/resource use and resource use for at least some levels of low rates of resource use. Below we provide evidence supporting this assertion from fieldbased estimates of production and resource use in Eucalyptus plantations, and from the emergent patterns in the 3-PG model of Landsberg and Waring (1997) Evidence for increasing resource use efficiency as forests increase rates of resource use Interest in resource use efficiency among ecologists was stimulated by Vitousek’s (1982) examination of nitrogen use efficiency as a function of nitrogen supply. Information on the N content of annual litterfall was available for many forests, and Vitousek (1982) used the mass of litterfall as an index of total forest growth, and the N content of litterfall as an index of N supply (nitrogen use efficiency = litterfall mass/litterfall N content). He plotted the ratio of litterfall mass:N as a function of N content of litterfall, and found that the efficiency of N use appeared to decline markedly as the supply of N increased. This approach could suffer from autocorrelation of the two axes (as the N content of litterfall appears in both), but Vitousek showed that the decline in N use efficiency differed somewhat from the trend that would result from autocorrelation alone (see also Pastor and Bridgham’s theoretical consideration of this trend). This approach would not work if litterfall mass was not a constant proportion of ecosystem production as ecosystem production increases, which may be the case for Eucalyptus plantations (and other forests?). The 14-site rainfall gradient examined by Stape et al. (2003a, this volume) showed the litterfall mass/N uptake pattern followed the declining efficiency pattern, whereas the actual trend in ANPP/N uptake showed the opposite (increasing) trend because of the striking decline of litterfall mass as a proportion of ANPP as ANPP increased (Figure 3). Goncalves et al. (1997) examined the pattern between annual aboveground increments of biomass and N, and found a linear increase (N uptake in kg ha yr = 1.87 + 3.1 times biomass increment in Mg ha yr, r = 0.84). The near-zero intercept and linear trend yielded a constant rate of wood production per unit of N in wood. Across gradients of forest production, wood production tends to increase more than leaf production, so we suspect this constant rate of wood production/N in wood may have resulted from an actual increase in the overall efficiency of production/N use by the trees. These studies clearly indicate that we should not necessarily expect the efficiency of N use to decline as N supply increases. Why would forests remain on the linear part of the slope of production versus resource use, rather than proceed farther into a region of declining marginal gains from marginal increases in resource use? The short answer might be that plants are integrated systems where investments in tissues (leaves and roots) to obtain resources (light, water, nutrients) are balanced to a greater or lesser degree, and “luxury” consumption of resources may not be not widespread in forests. In addition, changes in the rate of use of one resource is often associated with changes in the rate of use of other resources, and the efficiency of use of these other resources. Trees with greater access to water may demonstrate greater efficiencies of use for light and nitrogen than trees that are more drought-stressed. Some economic analogies have been developed by ecologists to gain insights on likely patterns of biomass allocation and nutrient use efficiencies (Bloom et al. 1985, Chapin et al. 2002). A full review of these ideas is beyond this paper, but we note that the definitions of economic analogs by Bloom et al. (1985) reduced the insights they could obtain from economic analogies (as described by Hof et al. 1990). For example, Bloom et al. (1985) used resources as analogs of economic inputs, and biomass as an analog of both economic products and economic revenues. Hof et al. (1990) recommended using the carbon (C) invested in roots and leaves as inputs, the resources obtained by leaves and roots as analogs of products, and the C gain through the use of these resources as revenue. Hof et al. (1990) concluded that the efficiency of using a resource (GPP/resource used) such as water should increase as the amount of water captured increased, not as a result of more efficient photosynthesis, but as a result of reduced leaf area. For the given amount of available light, trees with more water required less investment in roots to meet the water loss associated with a CO2 uptake, yielding an overall increase in net C gain per unit water used. As the inputs to a factory increase in availability, manufacturing becomes more efficient per unit of resource used because the inputs are cheaper to obtain; a direct analogy back to the discussion above of compensation points (or Rmin). Light use efficiency in wood production by Eucalyptus nitens A case study with Eucalyptus nitens from Australia illustrates the increased efficiency of using light to produce wood (wood production/light used) as the amount light interception increases. Smethurst et al. (2002) examined the overall pattern between leaf area index (LAI) and stemwood production across a range of stand treatments for 4 sites (Figure 4). Wood production increased as LAI increased from 2 to 10, with the greatest increase in wood growth/LAI occurring at an LAI of about 4. This would seem to indicate a declining efficiency of resource use with LAI > 4, but resource use does not relate linearly with LAI. Light capture per unit of leaf area declines exponentially as LAI increases (the Beer-law pattern), and conversion of LAI to light interception showed a sustained increase in wood production per unit light interception, and increasing efficiency of light use. For this conversion, we assumed a Beer-law light extinction coefficient of -0.5 and an annual light supply (photosynthetically active radiation, PAR) of 2.5 GJ/m, but the overall trend of increasing efficiency is consistent for any reasonable choice of parameters. Referring back to the production ecology equation, this increased efficiency of light use to produce wood could result from an overall increase in efficiency of light use, or from a shift in biomass allocation that covaries with overall forest growth. These two alternatives cannot be evaluated without more information, but the overall conclusion remains robust: wood production per unit of light intercepted increased with increasing light interception. Resource use efficiency and GPP for a plantation of Eucalyptus grandis x urophylla Stape et al. (2003a) found increases in ANPP per unit resource used in clonal stands of Eucalyptus urophylla x grandis as the rate of resource use increased across a rainfall gradient in Bahia, Brazil. This increase could have resulted from increased efficiency of production (GPP/resources) as a function of resource use, or it may have been an indirect result of high-rainfall sites experiencing lower vapor pressure deficit (which would increase rates of photosynthesis per unit of water transpired). A stronger test of the direct role of resource use comes from an irrigation experiment in the same region, where the water supply varied by treatment without any substantial effect on VPD (Stape 2002). We use production data (Figure 1) from a single year (with a normal rainfall pattern that included a 3month dry period) to illustrate how resource use efficiencies changed in response to irrigation (Figure 5). Production per unit of light used increased with irrigation, with the greatest increase in efficiency for wood production (wood production/light used), and the least for GPP (GPP/light used). The disproportionate increase in efficiency for wood production resulted from an increased allocation of GPP to wood production (and lower proportional allocation belowground) in irrigated plots. The same trend was apparent for the efficiency of water use, and for the efficiency of N use (with N use defined simply as the N content of the canopy). The effect of irrigation on water use efficiency was smaller (30% in GPP/water, 70% increase in wood/water) than the effect on light or N use efficiency (58% in GPP/light, 110% for wood/N). The simultaneous increase in efficiency of using all three of these resources in response to an increase in the supply of just one resource (water) illustrates the interacting nature of these calculations of efficiency; an improved supply of one resource commonly increases the efficiency of using other resources (Nambiar and Brown 1997), and simulation models may be required for a full accounting of responses. Patterns in Light and Water Use Efficiency in the 3PG Model Stape et al. (2003b, this volume) parameterized the 3-PG model for plantations of Eucalyptus grandis x urophylla in Bahia, Brazil. The parameterized version of the model provided good representations of leaf area, resource use, and biomass production and allocation. This model was designed to integrate resource use and forest production, but it does not have an a priori expectation about the efficiency of resource use. We took a base-case parameterization for 3-PG, representing the Eucalyptus stand from Stape et al. (2003b) for the non-irrigated treatment, and the normal year of precipitation (which included a 3-month dry period). We then changed the model parameterization to increase rainfall by up to 1000 mm/yr, with the additional rainfall spread evenly through the year. A second set of simulations allowed vapor pressure increase when rainfall was increased, based on an empirical relationship for this site. With increasing rainfall, GPP/water use was almost constant, within a 3% range across a 1000 mm gradient in rainfall (Figure 6). Allowing the VPD to decrease in response to increasing rainfall allowed water use efficiency to increase by 21%. The increased use of water drove substantial increases in the efficiency of light use; each 100 mm increment in annual rainfall increased the efficiency of light use by about 5% (for baseline VPD) or 8% (with VPD decreasing with rainfall increment). These simulations did not include the change in cloudiness that would be associated with increasing precipitation (likely to be less than 10% reduction in incoming light at this location). Overall, the changes in efficiency were smaller in these simulations than those actually observed in response to irrigation (Figures 1 and 5) or year-to-year variations in weather (Stape 2002), so more investigation of these patterns with case studies and models should be very productive. Patterns in Light Use Efficiency for Eucalyptus Plantations around the World Several studies have documented patterns of aboveground net primary production (ANPP) of Eucalyptus plantations in relation to intercepted light (Figure 6). All these studies showed increasing rates of ANPP with increasing light capture, and positive Xintercepts (=positive Rmin, and negative Y intercepts), which dictated that the ANPP/light intercepted must increase with increasing light interception. We note that some of this increased efficiency probably resulted from shifts in allocation of GPP away from belowground production and into aboveground production, but for the two studies that estimated the entire GPP budget (Stape 2002, Ryan et al. 2003), the increase in efficiency was also true for GPP/light intercepted. It is possible that the apparent linear trends in Figure 7 could turn into curves that demonstrate declining marginal returns from light capture beyond 3 or 4 GJ m yr, but we have no observations from Eucalyptus plantations with such high rates. Implications for Managing Forest Production This view of the production ecology of forests has three main implications for how foresters and forest scientists think about forest growth. The first is that the basic production ecology equation can provide insight to any pattern of forest growth. Why is wood production higher on one soil type than another, or why did fertilization lead to a large growth response on one site but not another? The production ecology equation has heuristic value in structuring the possible answers to these (and related) questions. Second, a clear understanding of patterns of resource use efficiency across resource gradients is fundamental to explaining forest growth. Across a rainfall gradient in Brazil, the increase in ANPP with increasing rain resulted more from the effects of water supply and VPD on the efficiency of water use than on the simple increase of water use (Stape et al. 2003a, this volume). The third implication is that environmental issues about the impacts of forest management hinge in part on resource use and efficiency of use. The rate of wood growth per m of water transpired by a forest varies greatly among sites; a given amount of water use can yield twice the wood increment on sites with higher water supplies and lower VPD. We look forward to studies that examine the ideas presented here in more detail, with stronger empirical tests of the patterns that have begun to emerge from recent experiments. We expect the overall trend toward increasing efficiency of resource use with increasing resource use (at the scale of forests and years) will remain robust, but the details, magnitudes and interactions among resources and biomass partitioning require much more work to provide a general picture of trends and exceptions. Acknowledgements. D.B. thanks John Aber for opening a door on why the efficiency of resource use is worth thinking about, and John Pastor, John Hof, and Doug Rideout for clarifying thoughts. This paper was supported in part by McIntire-Stennis appropriations to Colorado State University; J.L. Stape’s collaboration was supported by the CNPq Conselho Nacional de Desenvolvimento Cientifico e Tecnologico, of the Ministry for Science and Technology of Brazil (201144/97-2).