Marginal Land Produces Marginal Biomass

The global potential for energy crops is a topic of great interest, and the media is often filled with reports of the potential for production on marginal land. Indeed, some of these reports go so far as to suggest that a substantial fraction or even all of current global oil consumption could be replaced by energy crops grown on marginal soil. A new study was just released that makes such a claim:

Study Estimates Land Available for Biofuel Crops

Using detailed land analysis, Illinois researchers have found that biofuel crops cultivated on available land could produce up to half of the world’s current fuel consumption — without affecting food crops or pastureland.

The news release goes into some of the details of the study:

Using fuzzy logic modeling, a technique to address uncertainty and ambiguity in analysis, the researchers considered multiple scenarios for land availability. First, they considered only idle land and vegetation land with marginal productivity; for the second scenario, they added degraded or low-quality cropland. For the second scenario, they estimated 702 million hectares of land available for second-generation biofuel crops, such as switchgrass or miscanthus.

The researchers then expanded their sights to marginal grassland. A class of biofuel crops called low-impact high-diversity (LIHD) perennial grasses could produce bioenergy while maintaining grassland. While they have a lower ethanol yield than grasses such as miscanthus or switchgrass, LIHD grasses have minimal environmental impact and are similar to grassland’s natural land cover.

Adding LIHD crops grown on marginal grassland to the marginal cropland estimate from earlier scenarios nearly doubled the estimated land area to 1,107 million hectares* globally, even after subtracting possible pasture land — an area that would produce 26 to 56 percent of the world’s current liquid fuel consumption.

Speaking as someone who has made similar estimates, let me offer the following caveats. First, I agree with their assessment of available land area. I also attempted to estimate the amount of land that could potentially be available for energy crops in the chapter I wrote for Biofuels, Solar and Wind as Renewable Energy Systems. According to the The World Fact Book, there are 14.9 billion hectares of land area in the world, 13.31% of which are considered to be arable (but much of it marginal). Permanent crops occupy 4.71% of the total land area, leaving 12.8 million square kilometers (1.28 billion hectares) of arable land potentially available for cultivation of energy crops.

It is important to distinguish non-arable land (which people sometimes mistakenly refer to as marginal land) from marginally arable land (marginal land). Non-arable land falls into categories of desert, polar regions, some mountainous areas, etc. The entire continent of Antarctica is an example of non-arable land, encompassing 1.4 billion hectares. But marginal or even non-arable land can sometimes be made arable with irrigation or by applying soil nutrients.

The problem with using marginal land for energy crop production is that the land is deemed marginal for a reason. Marginal means that it simply won’t produce biomass like good arable land. This is the error many people make when trying to determine how much biomass can be grown on marginal land. They use yields more typical of arable land. I worked out an example in the aforementioned book chapter that illustrates how difficult it will be to replace today’s oil production with energy crops:

Consider how much petroleum might be displaced if all 1.28 billion hectares of arable land were planted in an energy crop with an oil productivity similar to rapeseed. While the average worldwide yield is substantially lower, rapeseed growers in Germany have succeeded in pushing oil yields to 2.9 tons/ha (Puppan 2002). If the rest of the world could achieve these high levels, this would result in a hypothetical worldwide oil yield of 3.7 billion tons. The energy content of rapeseed oil is about 10% less than that of petroleum diesel, so the gross petroleum equivalent yield from this exercise is 3.3 billion tons per year.

Because it takes energy to produce the biomass and process into fuel, the net yield will be lower, and in some cases may even be negative (i.e., more energy put into the process than is contained in the final product). Lewis compared several studies that examined the energy inputs required to produce biodiesel from rapeseed (Lewis 1997). Depending on the assumptions made, the energy input estimates ranged from 0.382 to 0.870 joules of input per joule of biodiesel produced and distributed. Assuming the best case value (lowest energy inputs) of 0.382, the net petroleum equivalent yield of rapeseed oil is reduced to 2 billion tons per year. The world’s present usage of petroleum, 85 million barrels per day, is equivalent to 4.25 billion metric tons per year.

Bear in mind that in the thought experiment above, I have presumed usage of all the available arable land in the world, used the best yields from good arable land, pretended that we could get those same yields from marginal land, assumed very low energy inputs to process the biomass into fuel — and still came up with less than half of today’s global oil consumption. I did this particular exercise to frame the global potential for biofuels, not knowing whether the result of the thought experiment would be a multiple of today’s oil consumption or a fraction of it. Because it was only a fraction, and because realistically the yields from marginal land are not likely to be a quarter of the best yields on arable land, I concluded that it was unlikely that biofuels could supply 10% of today’s oil demand.

Another caveat is that these studies often make unwarranted assumptions about the status of biofuel technologies. Even on the best land today, there are few biofuel crops that contribute to global energy supplies. Cellulosic ethanol has been out of commercial reach for 100 years, and yet this is the sort of presumed technology that will turn energy crops into the equivalent of half the world’s oil supply.

Finally, it is very important to remember that studies such as these are based on models. Models can be very important tools, but a model isn’t reality. In fact, models may not remotely resemble reality. I have used models throughout my career, and the most import part of building and using a model is model validation. That is the step where you feed the model data and see if it accurately reproduces known data.

For example, if I model a chemical process, I will validate that model by putting in actual process conditions and then checking whether the output corresponds to the actual plant output. Even then, you can’t be sure that varying the model will give results that will ultimately reflect reality, but without the step of model validation there can be no confidence at all in the model’s predictions

So the question I would ask of this study’s authors is: How did you validate your model? My guess is that it wasn’t validated, but the way it could be validated is to take some marginal land, produce some energy crops, turn them into fuel, and see if the predicted biofuel yields match what is actually demonstrated. My expectation is that it won’t, which means this is just one more study that lulls people into the false belief that on the other side of the oil depletion curve is a vast potential for biofuels just waiting to be scaled up.


Lewis, C. (1997). Fuel and Energy Production Emission Factors. MEET Project: Methodologies for Estimating Air Pollutant Emissions from Transport.

Puppan, D. (2002). Environmental evaluation of biofuels, Period Polytech Ser Soc Man Sci, 10, 95–116.

* One hectare is 2.5 acres.