全球变化知识资源中心

Using biomass to provide energy services is a strategically important option for increasing the global uptake of renewable energy. Yet the practicalities of accelerating deployment are mired in controversy over the potential resource conflicts that might occur, particularly over land, water and biodiversity conservation. This calls into question whether policies to promote bioenergy are justified. Here we examine the assumptions on which global bioenergy resource estimates are predicated. We find that there is a disjunct between the evidence that global bioenergy studies can provide and policymakers' desire for estimates that can straightforwardly guide policy targets. We highlight the need for bottom-up assessments informed by empirical studies, experimentation and cross-disciplinary learning to better inform the policy debate.

The large-scale production of renewable heat, electricity and transport fuel from biomass is an important component in many climate change mitigation and energy supply scenarios^{1, 2, 3, 4}. The International Energy Agency, for example, estimates that biomass could contribute an additional 50 EJ (~10%) to global primary energy supply by 2035, and states that “the potential supply could be an order of magnitude higher”⁴. Governments of the world's largest economies have also introduced policies to incentivize bioenergy deployment, motivated by concerns about energy security and climate change, and by the desire to stimulate rural development^{5, 6}. Yet the potential contribution from biomass to global energy supply is controversial. Sources of contention include concern about the interlinks between biomass, bioenergy and other systems. Most notably, land and resource conflicts are foreseen between bioenergy and food supply, water use and biodiversity conservation. The fear is that the benefits offered by increased biomass use will be outweighed by the costs^{7, 8, 9, 10}. It is also argued that the wide range of estimates of biomass potential and the lack of standardized assessment methodologies confuses policymakers, impedes effective action and fosters uncertainty and ambivalence¹¹. These broad points contribute to a general sense of unease about the future role of bioenergy, and whether it presents a genuine opportunity or is a utopian (or for some dystopian) vision that stands little chance of being realized.

Here, we analyse how scenarios for increasing bioenergy deployment are contingent on anticipated demand for food, energy and environmental protection, and expectations for technological advances. We use a systematic review methodology^{12, 13} to identify and analyse the most influential estimates of the global bioenergy potential that have been published over the past 20 years. The technical and sustainability assumptions that lie behind these estimates are exposed and their influence on calculations of potential is described.

We find that the range of estimates is primarily driven by the choice of alternative assumptions and that estimates should be viewed as 'what if' scenarios rather than forecasts or predictions. Larger estimates, however, are invariably based on more challenging assumptions, which would be more difficult to implement in practice.

The most controversial and influential assumptions relate to the future role of energy crops. We examine these assumptions, focusing on yield predictions, water availability and sustainability assurance. We find that studies provide limited insight into the level of deployment that might be achievable in practice and this highlights the need for caution in using global estimates to justify political intervention.

Finally, we highlight the need for better evidence, and recommend adopting a learning-by-doing approach to testing the feasibility and sustainability impact of increasing bioenergy deployment.

The global availability of biomass cannot be measured directly, it can only be modelled. Models vary in complexity and sophistication, but all aim to integrate information — from sources such as the Food and Agriculture Organization's (FAO) databases, field trials, satellite imaging data and demand predictions for energy, food, timber and other land-based products — to elucidate bioenergy's future role. The least complex approaches use simple rules and judgment to estimate the future share of land and residue streams available for bioenergy. The most complex use integrated assessment models that allow several variables and trade-offs to be analysed.

Although models differ greatly in scope and sophistication, the future supply of biomass in all cases depends on the availability (and productivity) of land for energy crops and food, and the ready supply of residues and wastes from existing and anticipated economic activity. Land availability is strongly influenced by assumptions about the area that should be set aside for nature conservation, along with population and diet scenarios — a vegetarian diet, for instance, requires less land than one rich in meat and dairy. Land productivity is affected by technology scenarios. Particularly important is the potential to increase crop yields and close the gap between optimal yields and those achieved by farmers when faced with environmental constraints such as water and nutrient scarcity, soil degradation and climate change^{14, 15, 16}.

Modelling results are most often discussed in terms of a hierarchy of potentials: theoretical > technical/geographic > economic > realistic/implementable. These terms are not always used consistently, and so results for different studies need to be normalized before they can be compared. Here, we compare estimates on the basis of the gross energy content of the biomass (assuming a calorific value of 18 GJ per oven dry tonne (odt)) and the chief technical and environmental assumptions on which they are predicated.

Our systematic review identified 90 studies. Of these, 28 contained original analyses describing over 120 estimates for the future contribution of biomass to global energy supply^{1, 14, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41}. Most of these estimates are for 2050, reflecting the importance of this date in much of the modelling and scenario analyses that have been done over the last 10 years. A detailed analysis of these studies provides the evidence base for this Review (see Supplementary Tables 1–4).

The most important potential sources of biomass are energy crops (22–1,272 EJ), agricultural residues (10–66 EJ), forestry residues (3–35 EJ), wastes (12–120 EJ) and forestry (60–230 EJ), summarized in Fig. 1. Not all studies include all of these categories in their analysis — in particular, many authors exclude biomass extraction from primary forests because they consider that the risk of adverse impacts on biodiversity and carbon stocks is too great. By way of comparison, the total human appropriation of net terrestrial primary production (including the entirety of global agriculture and commercial forestry) is around 320 EJ, of which 220 EJ is consumed and 100 EJ discarded as residues or otherwise destroyed during harvest⁴². This is considerably less than the current global primary energy supply (~550 EJ).

Vertical lines show the range of estimates for each resource category and diamonds indicate the results of individual studies (estimates include unconstrained values). Surplus agricultural land includes good quality land released from food production because yield growth exceeds demand (also called abandoned land in some studies). Rest land includes savannah, extensive grassland and shrubland. Degraded land may also be defined as low productivity or marginal land. Land categories cannot be considered fully mutually exclusive. Waste includes dung, municipal and industrial waste. Forestry describes harvest of a fraction of the global annual forest growth increment, and is a highly aggregate category defined by the FAO as areas spanning more than 0.5 ha with trees taller than 5 m. Some studies make further distinctions between primary forests and plantations.

Biomass potential estimates can be broadly divided into those that test the boundaries of what might be physically possible, and those that explore the boundaries of what might be socially acceptable or environmentally responsible. Through a detailed examination of each estimate we have identified the key assumptions that determine why bioenergy resource modellers reach such dramatically different conclusions. We describe the most important combinations of assumptions below, and they are summarized in Fig. 2.

In each band the minimum essential assumptions that must be included in global biomass models to achieve the given range of biomass potential are indicated. 'All residues' includes: wastes (dung, municipal and industrial), agricultural residues and forestry residues. Indicated global net primary productivity is aboveground terrestrial productivity only. Figure reproduced with permission from ref. 13, © 2011 Raphael Slade.

All studies of biomass potential assume that food demand will be met. How much land is needed is strongly influenced by yield projections for cereal crops. Cereals are of primary importance because about two thirds of all the energy in human diets is provided by just three crops — wheat, rice and maize⁴⁷ — which together already occupy 10% of the global land area. The main source of yield projections used in biomass studies to date is the FAO, and in particular two reports (published in 2003 and updated in 2006)^{48, 49} that describe yield growth for the major cereal crops increasing more or less linearly at 0.9% per year to 2050 (0.9–1.4% per year between 1999–2030; 0.5–0.7% per year between 2030–2050; compared with 1.6% per year for the period 1967–1999). There is concern, however, that these projections may be over-optimistic and give the impression that there is greater scope for productivity increases than is actually the case. Erb et al. identify that biologists tend to be among the most sceptical²¹.

The FAO's analysis was undertaken before the 2007/2008 commodity price spikes and one of the background assumptions in the 2003 report was that oil would cost less than US$30 per barrel and decrease to US$21 per barrel by 2015. In this scenario the cost of energy provides no constraint on agricultural production. Post 2007/2008, concern about rapidly rising prices rekindled interest in food security and spawned a series of influential reviews examining whether increasing food yields could meet the demands of a growing population^{50, 51, 52, 53, 54, 55, 56}. The FAO also updated their analysis, concluding that cereal yield increases of 0.9% per year to 2050 remain possible, but only if sufficient investment is forthcoming⁵⁷. The broad consensus of these reports was that it is likely to be technically possible to produce sufficient food to feed the 2050 global population, but there will be no room for complacency — particularly if the environmental impacts of global agriculture are also to be mitigated.

Yet these studies also highlight the inherent difficulties in undertaking a discussion about the world's capacity to produce sufficient food in abstract and aggregate terms. Digging beneath the surface of these analyses indicates that many of the underpinning assumptions are uncertain, in some cases contested, contingent on favourable investment scenarios and low energy prices, or subject to large regional variations. Rates of technological innovation and improvement are particularly problematic to anticipate as small changes make a big difference when compounded over several years in highly aggregate models. Focusing solely on the scope to increase food production also ignores issues such as post-harvest losses, food wastage and inequities in distribution⁵⁸. There are nevertheless some broad insights that might reasonably influence our interpretation of the bioenergy literature. First, the green revolution led to food production outpacing demand but at a major cost to the environment, and with greatly increased energy, water and nutrient inputs⁵⁹. Second, there is an opportunity to increase yields and close the gap between what farmers now get and what they might get with optimum agronomy, but many of the easy gains have already been achieved. The practicality of closing yield gaps is also hotly contested, varies dramatically by region and depends as much on political and institutional factors as it does on fundamental agronomy and the availability of nutrient and water inputs. Third, agricultural intensification is considered probable and necessary, but far from being a panacea it could further jeopardize the long-term sustainability of food production unless combined with measures to conserve and maintain soil fertility.

A critical assumption embodied in many bioenergy models is that as agricultural yields increase, crop and pasture land will be spared from production and can be made available for growing energy crops. The reasoning is that as yields increase, prices drop and the agricultural area will decline. This causal chain assumes that demand for the products does not change, and so the drop in price is sufficient to motivate land abandonment. If demand is elastic, however, prices may not change significantly. In this case the farmer has no incentive to abandon land, but may, conversely, be incentivized to increase the area they cultivate as this will directly lead to an increase in income⁶⁰. Empirical studies undertaken at local and regional levels provide evidence of both land-consuming and land-sparing effects from intensification, but a lack of robust data on abandoned land, as well as the confounding effects of global trade and political intervention, makes examining global level effect difficult^{61, 62}. Looking at changes in the global cultivated arable areas between 1970 and 2005, intensification only seems to be correlated with declines in cultivated areas between 1980–1985 in the aftermath of a sustained decline in agricultural commodity prices and a steep rise in yields⁶⁰. Moreover, explicit political intervention seems to have been an essential driver for cropland abandonment. There is some evidence that the developing countries that increased staple crop yields most rapidly in the period 1979–1999 had a slower deforestation rate than might otherwise have been the case⁶¹, but the overall conclusion is that the link between crop intensification and land sparing is weak and uncertain. It follows that bioenergy estimates that are contingent on land sparing — that is, those estimates in excess of ~300 EJ — must be considered at least as uncertain, if not more so.

This discussion suggests that where bioenergy models are based on aggregate productivity projections for food crops they must be interpreted with great caution. Bioenergy models can identify the most important relationships, for example the link between increasing meat consumption and demand for land, but the outputs are essentially 'what if' scenarios that possess no predictive capability and only hint at the level of effort that would be required to implement them. This is a striking contrast to the International Energy Agency's high expectations for an additional 50 EJ contribution to primary energy by 2035.

Globally, agriculture accounts for ~70% of all fresh-water use, and scarcity is a growing concern⁶³. The vast majority of this water is consumed during crop cultivation: either evaporated from the soil or transpired from plant leaves^{63, 64}. Yield and water transpiration are closely correlated and maximum crop growth only takes place when water availability is not restricted⁶⁵. Crop growth models are able to predict water-restricted yields for both food and energy crops, but competing demands on water supplies are not considered in depth in global bioenergy studies. A few irrigated energy crop scenarios have been developed for illustrative purposes, however the authors c