Dairy Queen here: This is the second guest post from our Berkeley neighbor Marc (bio now available), who's also been contributing daily to the Digest. Here, Marc — who's an engineer — engages in a little recreational fact-checking that I for one found fascinating.
[Updated March 20 with caveat on Lundberg Rice and photo credit]
Local vs. imported rice: Doing the math
One of the few controversial topics at the recent discussion between Michael Pollan and Whole Foods founder and CEO John Mackey was whether local was always the better choice for conscious consumers — whether "food miles," or the distance that food travels from field to plate, should outweigh other factors in production. But when Mackey asserted that Berkeley residents concerned about their carbon footprint should buy rice from Bangladesh instead of that grown in the state, the audience grumbled.
"Hey, if you don't like it, talk to Peter Singer," he said. He was referring to the ethicist's latest book, The Way We Eat (co-written with Jim Mason), which argues that rice imported from Bangladesh requires less energy in total and leads to lower carbon dioxide emissions than California-grown rice. Mackey recommended that skeptics check out Singer and Mason's methodology on their own.
I took Mackey's advice (after a bit of inspiration from Twilight Greenaway's post at Culinate) and reviewed Singer and Mason's arguments. On page 148 of the cloth edition, they write that "Rice is grown in California, under irrigation, but it takes a lot of energy to grow it there — about 15 to 25 times as much energy as it takes to grow rice by low-energy input methods in Bangladesh." The energy comparison comes from footnote 32, a report by Jules Pretty and Andrew Ball, two researchers at the University of Essex (Ref. 1 in the list at the bottom of the post). Pretty and Ball give an energy input for Bangladeshi "traditional deepwater rice" of 400 megajoules per metric ton (a joule is a unit of energy, and a megajoule, abbreviated as MJ, is 1 million joules; a metric ton is 1,000 kg) and 11,000 MJ per ton for California rice. I'll return to Singer and Mason at the end of the post.
A new look at energy
I performed my own literature search, compiled the data from peer-reviewed journals and public documents, then made some simple calculations to find the total energy required. As a practical illustration, I compare the energy required to produce and transport two 2.5 kg bags of rice (2.5 kg = 5.5 lb, 0.0025 metric tons), one from Bangladesh, the other from Richvale, California (home of Lundberg Farms, one of California's premier producers of long-grain rice. I chose Lundberg as an example because they were the only California long-grain rice producer I could think of, not because they are an example of high-input industrial agriculture In contrast, Lundberg is one of the pioneers in organic and low-input rice farming in the United States, so their energy inputs are probably less than average.). The details are after the jump, but I'll give my conclusion before the jump to those who can't wait — or have a fear of math and energy calculations.
My calculations were inconclusive as to which rice requires more energy, but the two sources were reasonably close. It requires about 14 times more energy to move the rice from Asia via container ship than from Richvale via truck, but significantly more energy is used to grow and process rice in California than in Bangladesh. However, my source for Bangladeshi energy inputs shows a far higher energy input than Singer and Mason cite. I used two different acceptable estimates for California rice energy, with one of them resulting in the California rice requiring 12 percent less energy for production and transportation, the other resulting in 18 percent more energy.
The energy calculation
I found a 2005 paper in the American Journal of Environmental Sciences by three researchers from Bangladesh that provides estimates of the total Bangladeshi agricultural energy input (Ref. 2). They collect data on human power, animal power, machines, fertilizer, irrigation, and so forth using government reports, field studies, and simple calculations.
The energy input devoted to rice is not specifically listed. Because 75% of the agricultural land is for rice production, I assumed that the total agricultural input on a per hectare basis (one hectare = 2.47 acres, abbreviated as ha) was reasonable to use for rice. In 2001, the yield was 2.81 metric tons of milled rice per hectare. (The paper provides yield in terms of rough rice, I used the Ref. 3 figure of a 29 percent mass loss, i.e., 1 kg of rough rice is 0.71 kg after milling.) Energy consumption was 17.43 gigajoules per hectare (a gigajoule is 1 billion joules, abbreviated as GJ).
Thus, 6.2 GJ are needed to produce a ton of milled rice in Bangladesh, and 15.5 MJ for 2.5 kg of rice.
In this calculation, I'm interested in the use of nonrenewable energy, such as diesel fuel or chemical fertilizer. However, the paper does not specify how much of the energy input is fossil energy. About 12 percent of the total energy input is human, and about 2 percent is animal power, two forms of energy input that are not completely fossil-fuel-based. For example, some of the energy supplied by draft animals comes from naturally growing vegetation, which requires little fossil fuel input, if any. Similarly, only a portion of the energy used to grow food to generate human power is fossil energy (i.e., it requires less than one calorie of fossil energy to create a calorie of food for a human). Determining how much human and animal energy is derived from fossil fuel is one of the many difficulties of an analysis of this sort.
I assume that the rice is transported by a 1,500 TEU container ship. A "TEU" is a 20-foot container, so this ship carries the equivalent of 1,500 20-foot shipping containers from the Port of Chittagong, Bangladesh, to the Port of Oakland, California, an over-water distance of 14,920 km (from Port World Calculator). A paper by Kristensen in the journal Marine Technology (Ref. 4) provides energy consumption figures in terms of the number of megajoules used to carry a ton of cargo a distance of one kilometer. For a 1,500 TEU container ship, the energy usage is 0.165 megajoules per ton per km, or 5.96 MJ to carry a 2.5 kg bag of rice from Bangladesh to Oakland. The largest ship in the Kristensen dataset (6,000 TEU) uses 13 percent less energy per ton-km than the 1,500 TEU ship.
My calculation is missing some inputs, like transportation from rice fields to the port of Chittagong, and energy consumption at the both ports (most goods are moved many times before loading and after unloading), but should give a reasonable estimate.
In California, the yield of milled long-grain rice in 2001 was 6.18 metric tons per hectare (Table 9 in Ref. 3 below). The energy input is 39.5 GJ per hectare, from a 1980 book about energy use in agriculture (Ref. 5 below). I did not include the line items for "drying" and "transportation" because the Bangladesh study did not mention those two activities, for a reduction of 6.3 GJ/ha from the book's total of 45.8 GJ/ha. Thus, for 2.5 kg of milled rice, the energy required is 16 MJ. In the summary, I'll call this "California A."
As an alternate method, I calculated the energy input using a combination of the ratio of output energy to input energy from Ref. 5 (1.76) and the yield from Ref. 3 (6.18 t/ha) to obtain an estimate of 24.8 MJ for 2.5 kg of milled rice. I'll call this energy estimate "California B."
Almost all of the energy input for the California rice is non-biological. Only the "seed" line item is a biological input, accounting for only 7% of the energy in the estimate. Human labor is such a small component that the authors didn't even include its energy component in the estimate (it is 23.6 hours per hectare).
Since most Bangladeshi rice is long grain, I picked a California farm that specializes in long-grain rice, Lundberg Farms in Richvale, California. Richvale is 152 road miles (246 km) from my main source of non-produce foods, Berkeley Bowl Marketplace (distance calculated from Google Maps' driving directions). The Kristensen paper (Ref. 4) gives an energy consumption for a heavy-duty truck of 0.69 MJ/ton/km, for 0.42 MJ to transport the 2.5 kg of rice a distance of 246 km.
Comparison with Singer and Mason's source
The energy input cited by Singer and Mason for Bangladesh is about 400 MJ/ton, a figure that is 15 times lower than the energy input in Ref. 2 (6,200 MJ/ton). For California rice, the energy input is 11,000 MJ/ton. The number that I derived from Ref. 3 and Ref. 5 was either 6,000 or 9,000 MJ/ton. The factor of fifteen difference caught my attention, so I attempted to find the original sources at the University of California library. I was able to find the two sources in Singer and Mason's footnote 32 (one a book, the other an on-line report), but the citations I could find were not about rice. My guess is that the 400 MJ/ton figure is from either a 1981 report by the International Rice Research Institute or a 1985 paper at a USAID conference. In the absence of the actual article, I can only speculate on why the Singer and Mason number is so low. My guess is that the number was used as a lower bound in discussions about how agriculture can become more sustainable through low-input organic farming methods. In a sense, it is a target to shoot for. The numbers in Ref. 2 are fairly recent, and reflect the evolution of farming since the "green revolution" reached Bangladesh (i.e., intensive use of chemical fertilizers and mechanization).
The two tables below summarize the results for the Bangladeshi and California rice with rows for production and transport. The first table is the total energy. The second is for "non-biological" energy, which excludes human power, animal power, and seed energy. In both the A and B estimates, California rice requires more energy to produce — 20 and 60 percent more. The Bangladeshi rice's ocean voyage requires about 14 times more energy than the California rice's truck trip. When considering total energy sources (biological and non-biological), the California A scenario's energy input is about 23 percent lower than the Bangladeshi scenario's input, but the California B estimate is about 18 percent higher. When considering non-biological energy sources only, California A is only 13 percent lower, while California B is 32 percent higher.
Energy inputs, all energy sources (in megajoules for 2.5 kg of rice)
|Bangladesh||California A||California B|
Energy inputs, non-biological energy sources only (in megajoules for 2.5 kg of rice)
|Bangladesh||California A||California B|
I must emphasize that the inputs have significant uncertainty — 27-year-old data for California rice, for example — and so these results should not be trumpeted as "truth." The point of this exercise was to explore how production and transport affect the overall energy consumption of one kind of food. I'd like to update this with better inputs, so if you know of any more recent sources of energy input to California rice production, please mention them in the comments. In addition, I was unable to find information about emissions of greenhouse gases during rice production (carbon dioxide, methane, and nitrous oxide are the ones related to agriculture and transportation).
Returning to Singer and Mason: They follow up their argument about imported and local rice with comments about energy consumption by the shopper. It doesn't take a very long car trip to burn up a few megajoules of gasoline energy, so driving a few miles out of the way to buy a local product will probably erase any energy benefit the local product brings. For example, gasoline contains 32 MJ per liter (source: U.S. EPA). A car that travels 12.8 km on a liter of gasoline (30 mpg) will use about 2.5 MJ per km (4 MJ/mi).
Energy is an important consideration, but certainly not the only one. Our food choices also need to consider how the land and animals are treated, how the food tastes, how the workers are treated, and how the food will affect our bodies.
1. Pretty, J. and Ball, A., Agricultural Influences on Carbon Emissions and Sequestration: A Review of Evidence and the Emerging Trading Options, Centre for Environment and Society Occasional Paper 2001-03, University of Essex, March 2001 (PDF)
2. Alam, M.S., Alam, M.R., and Islam, K.K., Energy Flow in Agriculture: Bangladesh, American Journal of Environmental Sciences 1 (3): 213-220, 2005 (first two authors are from Chittagong University of Engineering and Technology, last author is from the Islamic University of Technology, Guzipur). Full text in PDF
3. USDA Rice Situation and Outlook Yearbook, Market and Trade Economics Division, Economic Research Service, U.S. Department of Agriculture, November 2002, RCS-2002 (PDF)
4. Kristensen, H.O., Cargo Transport by Sea and Road — Technical and Economic Environmental Factors, Marine Technology, Vol. 39, No. 4, October 2002, pp. 239–249.
5. Rutger, J.N. and Grant, W.R., "Energy Use in Rice Production," in Handbook of Energy Utilization in Agriculture (D. Pimentel, Ed.), CRC Press (Boca Raton, FL), 1980. A snippet of the data from Arkansas can be found at a Purdue University site.
Photo credit: Photos from the USDA Image Gallery.