Sunday, September 2, 2018

Don’t let an LCA determine your menu

Dear readers, my "spring break" became a summer break, the warmest and driest summer on record for a very long time. But crops survived on our small farm thanks to the lake next to it. And we have harvested more than the feed we need for our 5 suckler cows and their offspring. Soon, I will write a post about the farm, but now: something totally different: Life cycle analysis.

Lately, life cycle analyses have become the predominant method for assessment of a product’s environmental impact in general and its climate impact in particular. We are exposed to various figures comparing beef to soy, organic to conventional, local to overseas etc. Publicity-hungry researchers and media spread those figures and conclusions all over the place. Very few make the effort to read the research articles and even less the supplementary materials which the articles are based upon.

At the moment I am conducting a comparison between milk and plant-based “milks” (the EU recently prohibited that most of them are marketed with the word milk, and the issue is under heavy debate in the US). The work has given me reason to dig deeper into a big pile of lifecycle analyses. Admittedly, I was already earlier quite skeptical to the method, or at least to how the method is used, and now I am more than skeptical. It is one thing for a company to use the method to assess different processes to make the same product, or possibly to compare two different products with the same function, it is something totally different when results of comparisons of totally different production systems are supposed to guide policy or consumer behavior.

The flaws and shortcomings appear on many levels. Some flaws are linked to the functional unit and how impacts is distributed among co-products. Many flaws emerge from the use of standard values and databases to assign impacts, i.e. in most cases the analysis doesn’t actually measure anything, just take data from other sources and feed into the spreadsheet. Even bigger shortcomings result from the lack of context of the analysis, little understanding of agriculture systems and even more ecological systems and how changes in production change consumption.

If agronomist makes the analysis they often use land area (hectare) as the functional unit.  But if you do little and harvest little, the environmental impact is likely to be minimal per area unit, which is fine, but also somewhat irrelevant. Extremely extensive grazing is (mostly) environmentally benign, but it also produce very little food. Food products are often analyzed per kilogram. But kg is not a meaningful unit when it comes to food. It is virtually impossible to feed a person with broccoli; you need to eat 10 kg per day to get your energy needs from it. You can choose calories as the measure, but then sugar will always be the most efficient product. Protein is perhaps a more relevant measure, but in some countries protein intakes are excessive, so it is not ideal either. Some try to make more sophisticated nutritional indexes. Smedman et al used such an index comparing milk with plant based milks. They estimated how many essential nutrients were covered by at least 5 % of the daily needs when 100 gram was consumed. Their research showed that milk had the least climate impact of all beverages tested. [i] Other researchers, however, contested the results and showed that if one used a 20 % threshold, orange juice was clearly the best and if one used a 2% threshold, soy milk was the best.[ii] There is no way one can state that one kind of functional unit is the Right one. That the choice of functional unit has such a big influence on the results is disturbing. 

How impacts are distributed among co-products is another important issue. If you analyze low fat milk for instance, you have butter or cream as an important co-product, so you have to determine how much of the impact should be allocated to milk and how much to butter or use some other method. But it goes much further than that. Each cow also gives birth to a calf every year, and those are either used as replacement, or killed at birth or raised for meat. And then the cow herself will sooner or later meet her Creator and be converted into hide, meat and other by-products. I will not expand on this technical matter here, but in the end how you allocate impacts has a major impact on your results.   

In agriculture, changes in land use, also poses major challenges. If soy bean production expand into the Amazon, and cause deforestation, carbon emissions and loss of biodiversity, is that impact a function of those particular fields, of soy bean production in Brazil or soy bean production in the world? In addition, there is also the question for how long the impacts should be distributed. Is it one year, twenty or hundred or even more? You can chose whichever method, there is no right or wrong, but the consequences of the choice is huge.

de Ruiter at al argue that as a result of global trade soy beans are interchangeable with many other crops, and therefore, changes in land use should be distributed on all arable land in the world.[iii] But it doesn’t stop there, recently published research demonstrate that the importation of soy beans into Europe has caused the abandonment of 6 million hectares of semi-natural pastures and meadows. [iv] Other research show how the large scale trade in feeds and grains result in accumulation of nutrients, such as phosphorus, and eutrophication in major importing regions, such as Europe. Such linkages are never part of lifecycle analyses.  

For climate impact, the role of methane and nitrous oxide are critical for the results, in particular for ruminant livestock. For ruminant products more than half of the emissions are caused by methane and about ¼ is from nitrous oxide (the exact composition depends on local conditions. It is fairly simple to estimate carbon dioxide emissions from the food chain as it is a direct result of the use of fossil fuels, where one atom of carbon in oil, coal or gas will become one molecule of CO2. For methane and nitrous oxide things are much more complex. Those conducting LCA’s are not measuring emissions of methane or nitrous oxide in the cases they study. They just put in data from other sources in spreadsheet models and out comes figures of how much nitrous oxide and methane are generated by the process. But there are huge uncertainties and variations in these emissions, in particular for nitrous oxide. Life cycle analysts use simple conversion factors (from IPCC or other databases or public sources) which translate the quantity of feed consumed, manure deposited or chemical fertilizers used into emissions. But scientific research show how uncertain these estimates are and that there are huge variations.

When Brazilian researchers actually measured how much nitrous oxide was emitted from the urination of cows on pasture, they found that merely 0.2% of the nitrogen was converted to nitrous oxide. This is one tenth of the standard emission factor used by the IPCC. A meta-analysis of 422 studies of nitrous oxide emissions from land fertilized by animal manure or chemical fertilizer revealed that emissions are considerably lower than the IPCC standard emission factors for manure and considerably higher for chemical fertilizer.

While also methane emissions can vary considerably, an even bigger problem is the conversion of methane emissions into carbon dioxide equivalents in order to get one figure of the climate impact. Through this conversion the real climate impact of methane is tremendously exaggerated.

LCAs don’t include effects in animal welfare and they also neglect social and cultural impacts, which if course could be even more important than environmental impacts. But LCAs also never include all environmental impacts. Effects on soil fertility and erosion are mostly left out as well as impact on bio-diversity. Even for categories that are often included, the indicators used are far from satisfactory, e.g. for impacts of pesticide use (it is obviously not possible to have one measure that will fully capture the impact of hundreds of pesticides in very different ecological conditions).

Variability between farming system as well as within the same system is seldom captured in single LCAs. This is also one of the reasons for why results differ so widely between different LCAs. For cow’s milk the land use range from 0.7 m2 to 242 m2 per liter milk according to the meta-analysis by Poore and Nemecek recently published in Science.[v] If an LCA is made on the farm level for oats the results will be very different between a year with a bumper yield compared to a poor year. As a matter of fact, it is not uncommon that the whole crop of a certain farm is classified as not suitable for human consumption a bad year. What happens then to the LCA results of oat milk?


But the major inadequacy of LCAs is that they don’t cover system effects and effects outside of the boundaries of the analysis. A general rule of ecology formulated by Garret Hardin is “you can never change one thing”, because as soon as you do that, other things change as well. In the example of “milk”. If you were to replace cow’s milk with soy milk in Sweden, the whole farming system in Sweden would change and the impacts on landscape, rural economy, crop rotations etc. would be much bigger than you could read from a lifecycle analysis. In addition, as you can’t produce soy in Sweden it would trigger expansion of soy production in other parts of the world, thereby contributing to deforestation, increased use of pesticides etc. If oat milk is the substitute you will again get other results.     

The most important changes of the agricultural system in Western Europe over the past seventy years is the abandonment of small farms, small fields, pastures and marginal lands and the decreasing diversity of the farming systems. There are fewer farms and each farm mechanize and specialize into a few crops or one kind of animals. Not only farms specialize, whole regions do, even countries do. Livestock and crop production has been separated and cycles of nutrients between animals and crops as well as between farms and the rest of the food system have been broken. Largely this has been ”solved” by massive application of chemical fertilizers. The simplification of the production system has also led to the use of pesticides. These massive changes in the agriculture landscape are not captured by LCAs, and the drivers causing this are also not understood by reading hundreds of life cycle analysis.

In general, LCAs contribute little to the understanding of agriculture systems; on the contrary, when looking into LCAs one can get many mistaken ideas. One prominent example is the comparison of livestock production and crop production. If they are compared side by side, product by product, almost all comparisons will conclude that it is much more efficient (from a multitude of indicators) to eat plants than livestock products. Subsequently professor Poore, who was one of the researchers behind the meta-analysis of LCAs, tells media that “A vegan diet is probably the single biggest way to reduce your impact on planet Earth, not just greenhouse gases, but global acidification, eutrophication, land use and water use.” But that is typical of the distorted view of reality you get from LCAs. Apart from the fact that livestock products have very favorable nutritional profile, which was not part of the analysis, some livestock production actually make use of crop products which would not otherwise be used.

A considerable part of the crops are not edible by humans. Almost all oil crops have protein rich by-products which can’t be eaten by people. Already now these oil cakes are major animal feed and with no livestock products vegetable oil production must increase even more (most likely this means even more palm oil). And there will be more leftovers which could have been fed to animals. Major grains, such as wheat and rice have big shares of non-edible by-products; for wheat, normally only 80 % of the grain harvest goes to flour and for rice it is even less. Then there are leftovers from the production of sugar, beer and liquor, even the iconic tofu and oat milk have by-products which are fed to animals. Add to this food waste, straw etc. you realize that a certain proportion animal products reduce resource use. In addition, some years, part of the wheat crop is not good enough for flour or pasta, some of the barley is not up to specs for malt and the oat is not good enough for oatmeal or oat milk. What happens with it? Animals eat it of course. Finally, when livestock is integrated with crop production rotations can be much better and yields of crops will benefit. In essence, the miraculous increase of productivity, and the end of food shortages in Europe 1800-1940 was largely driven by integration of livestock and crop production.

There are plenty of studies that clearly demonstrate that a certain fraction of livestock products is favourable. [vi] [vii] One can’t quantify how high proportion as this depends on climate, water and land resources. In addition, there are agro-ecological zones where crop production is very challenging, but where grazing lifestock can use the land to produce food. When the professor Poore tells media that livestock uses 83 % of the farm land and only produces 18 % of the calories, he conveniently omits that most of this land is far too dry, too steep or too cold for any crop production at all. 

Finally, people are largely missing from LCAs, which means that the changes in behavior which are triggered by new methods or new technologies are not captured. It is well known that if a product becomes cheaper, consumption will increase. Through the industrialization of agriculture crop production, food is very cheap, but this has also increased food waste tremendously. Chicken consumption has increased almost 10 times globally over the last fifty years, and this is largely driven by higher efficiency on chicken breeding, reducing feed use as well as labor use. Chicken is now one of the cheapest proteins you can buy. Chicken has not mainly crowded out more resource demanding beef and pork but rather the less resource-demanding grains and pulses, and in the process it has lost its role as small scale food residue converter. So, while the environmental foot print of chicken is small compared to beef according to LCAs, in reality, the changes in production and changes in markets and consumption has led to a huge increase in the size of the environmental foot print of human food chains. Other similar examples can be found with packaging and transports.

Many of those working with Lifecycle analysis are fully or partly aware of the limitations. Bruno Notarnicola and other LCA experts write in The role of life cycle assessment in supporting sustainable agri-food systems: A review of the challenges, in the  Journal of Cleaner Production that deficiencies in LCA methodology ”results in severe limitations when agricultural systems are being evaluated” and they caution that while the LCA concept is easy to understand and the results are user-friendly, most people, even many of those conducting LCAs, are interpreting the results in the wrong way without due consideration of assumptions and limitations.[viii]

Lifecycle analysis has its merits, but it is far from being the method that will give us easy answers to all the difficult questions there are about what we will have for lunch.

[i] Smedman, Annika et al 2010, Nutrient density of beverages in relation to climate impact, Food Nutr. Res. 54.

[ii] Scarborough, Peter och Mike Rayner 2010, Nutrient density to climate impact is an inappropriate system for ranking beverages in order of climate impact per nutritional value, Food Nutr. Res. 54. 

[iii] De Ruiter, H, m.fl. 2016, Global cropland and greenhouse gas impacts of UK food supply are increasingly located overseas, Journal of the Royal Society Interface Volume 13, issue 114

[iv] Boerema, Annelies m.fl. 2016, Soybean trade:Ballancing Environmental and Socio.Economic Impacts of an Intercontinental Market, PLOS One 31 maj 2016.

[v] Poore, J. och T. Nemecek 2018, Reducing food’s environmental impacts through producers and consumers, Science 360, 987–992 (2018).

[vi] Zanten, Hannah m.fl. Defining a land boundary for sustainable livestock consumption.

[vii] Peters, Christian J. m.fl. 2016, Carrying capacity of US agricultural land, ten diet scenarios, Elementa:Science of the Antropocene. 

[viii] Notarnicola, Bruno m.fl. 2017, The role of life cycle assessment in supporting sustainable agri-food systems: A review of the challenges, Jorunal of Cleaner Production 140 (2017) 399-409

No comments:

Post a Comment