Rod McDonald  

Email contact: info at sustainableprotein.com

Updated 8 July 2019

Leaves: an abundant source of protein

The protein in green leaves is abundant but is greatly under-utilized. Leaf protein concentrate extracted from lucerne is already on offer as a commercial aquaculture feed, but that process extracts only a small fraction of the protein that is in the leaves. An advanced process, developed to pilot scale in New Zealand, can extract 80% of the protein in common legumes and grasses. This process would allow a protein yield of over 2 tonnes per hectare, more than is obtained from conventional seed crops.

The basic products from the extraction process are: a solid protein/lipid concentrate that is 50% protein and is also rich in short-chain omega-3 fatty acids and antioxidants; finely divided low-lignin cellulosic fiber; and a liquid that contains sugars and other plant-cell nutrients. The fiber has potential for use as a bioenergy source to power the processing operations, perhaps with energy to spare for biofuel manufacture. The sugar-rich liquid should be useful for fermentation, perhaps for growing algae or yeast to produce long-chain omega-3 fatty acids to supplement aquaculture feeds. A variation on the process allows a higher concentration, higher digestibility protein fraction to be produced.

Selection and breeding for optimal yields of protein, lipids and other aquaculture feed ingredients from leaves has hardly been explored at all. The range of plant species or mixture of species that could be used is very wide, and even toxic plants might in some cases be detoxified during the processing. Compared with conventional monoculture crops, a production system based on perennial, deep-rooted leafy crops should be less vulnerable to catastrophic loss from pests, diseases, weeds, and the increasing weather instability that is expected to result from climate change.

This article is focused on producing feed for aquaculture, because that is a shorter step to acceptance than using it as a direct human food. However, the potential value for human nutrition, established since trials by Norman Pirie in the 1960s, is discussed near the end.



The aquaculture industry is working hard to substitute plant proteins and lipids for the traditional wild-caught fish in feed formulations, with considerable success. The “plant proteins” being used almost always come from arable seed crops, and mostly from soybeans.

Unfortunately, the expected changes in population, climate, and resources are likely to threaten the availability and reliability of arable seed crops. But in this presentation I will be describing a technology that could provide aquaculture with an alternative plant-based protein/lipid feed from non-arable land.

Leaf crops contain abundant protein, lipids and pigments.

As an alternative to annual seed crops, leaf crops have the advantages that they are usually perennial, they will be more resilient when challenged by droughts, floods and pests, and can better tolerate land that is marginal for arable crops[1].

The downside of leaf crops is that those nutrients are not readily accessible, being protected by tough cell walls. But those cells can be broken open by mechanical processing.

Obtaining aquaculture feed from leaf crops is not a new idea. A concentrated extract of lucerne (alfalfa), made in France, has been sold as an aquaculture feed ingredient for years under the name "Vitalfa" [2].  However, the process used provides a low yield of protein concentrate per hectare of harvest, and the asking price is high, similar to fishmeal. I have been told that price and lack of a large-scale source of supply excluded it from the European Aquamax trials that were aimed at finding alternatives to fishmeal[3].

The purpose of this presentation is to describe a different way of extracting nutrients from leaves, with a yield fourfold higher. It could be further extended to make long-chain omega-3 lipids as well.

Above on the left is a sketch of the long-established leaf protein extraction process. In France the lucerne crop is harvested 6-8 times a year while it is in a vegetative stage, when the leaves are lush and usually contain 25% protein or more.  

The protein concentrate yield is only 7% of the crop infeed, on a DM basis[4]. There is a good reason why the yield is low - the main product is dehydrated lucerne (the “Dehy”), and the protein concentrate is a only minor by-product. [5].

The dehydrated lucerne has to meet minimum protein-content specifications, so only a small proportion of the protein present in the leaves can be removed, 10-20%. The pulping and pressing operation that breaks open the plant cells and liberates the cell contents does not need to be very effective.

In the 1980s we developed a high-yield version of the process, which we developed, shown above on the right. By avoiding the production of dehydrated fibre, different milling techniques can be used to increase the extraction yield.

We intended that the fibrous fraction and the sugars would be used for biofuels. We demonstrated that with the right methods up to 80% and typically 70% of the protein can be extracted from ryegrass-clover mixtures and from lucerne[6], which were the only crops that we worked on.

The changes in the process do not look extensive, but the recoverable yield of protein per hectare per year can be four times that achieved by the low-yield process. The protein concentrate yield can be 28% of the crop dry weight instead of 7% in the older process.

In terms of protein per hectare, the yield would be over two tonnes, more than double that produced by a typical soybean crop[7] [8].

The quantity of lipids extracted along with the leaf protein would be similar to the quantity in a soybean crop, but with omega-3s predominant, where soy lipids are predominantly omega-6.

If the product is for aquaculture, the drying stage may be avoidable. The protein concentrate is already semi-dry when it emerges from the centrifuge.


Our low-yield process

We started in the 1970s with a low-extraction version not very different from the French approach, because we too were interested in dehydrated fibre as the main product.

After a few years of research involving agronomy, chemistry and animal nutrition as well as engineering, we went into partnership with two New Zealand companies to commercialize the process. We built a factory based on an existing lucerne dehydration plant at Reporoa, in the central North Island.

My team designed and supplied the pulping and pressing machinery, and made some innovations in the protein-separation part of the process.

When we started running the plant we had some learning experiences involving foam, and then the old dehydration plant burned down. But the company arose renewed from the ashes, thanks to insurance, and operated for another dozen years.

The pictures below show the first low-yield pilot plant; the pulping and pressing machines we built for the commercial plant;some processing problems; the old alfalfa-dehydration plant after it was destroyed by fire; the new design of dehydration plant; feeding the factory; and the product.


Development of the high-yield process

As soon as the operation was independent, the engineering team moved on to a high-yield version of the process, including cellulose-hydrolysis and fermentation work. That was the time of the 1980s liquid fuel crisis, and the government was encouraging biofuel research.

The pictures below show the first-stage pulper at top left and the high-intensity mill at top right, with a screw press beneath each. The added milling stage makes the fibre fraction much finer and lower in protein.


In the late 1980s the energy crisis disappeared. The research work was terminated and rapidly forgotten.



With the return of world concerns over the security of energy and food supplies, and the recent development of new biotechnologies, the process is worth re-considering. Of particular interest is how the process might integrate with the production of microbial lipids and proteins using algae or yeast.

Around the world there has been heavy investment in pilot projects for land-based production of algae, mainly for biofuel purposes but also for aquaculture feed. Some species of algae can be made to produce high concentrations of long-chain omega-3 fatty acids[9] or pigments[10] [11]  that are valued in aquaculture.

Algae can be produced in fermenters, without the aid of sunlight, and with sugars supplied from an external source, a method promoted by Solazyme™[12] and others. Under these conditions the algae can be grown much more densely than in solar ponds[13].

Algae can be difficult to culture economically, especially with a feedstock with variable composition, but yeasts are more robust. A yeast species has been modified to produce long-chain omega-3 fatty acids by fermentation, which was the subject of a  collaboration between DuPont™ and Aquaculture Chile[14]. The genetic modification in DuPont's yeast caused market resistance, but the use of omega-3s from micro-algae seems to have been accepted more readily [15].

Referring back to the high-yield process, something has to be done with the sugar-rich stream that remains when the protein and associated lipids are separated from the juice. We called that stream “deproteinized juice” or DPJ.

The DPJ contains about 15% of the crop on a dry-matter (DM) basis. It is a solution with about 5% DM content, of which some 30% is simple hexose sugars and another 20% is complex sugars like fructosan[16].

Fermentation processes require N, P and other trace nutrients as well as sugar. Nutrients are a significant cost and supply challenge for fermentation[17] [18]. The fact that DPJ is the contents of living plant cells, lacking only specific proteins, suggests that it should be a complete source of nutrients for yeasts without further additives, and therefore might have more value than the sugar-mill wastes that are likely to be the cheapest feedstock.

DPJ has been used successfully in various small-scale fermentation processes[19] [20] [21], so it seems reasonable to assume that the mixture would also be a good nutrient supply for algae, although I am unaware of anyone having tried it for that purpose.

If DPJ was used as a nutrient supply for an algae or yeast fermentation producing a high concentration of lipids, we could get this result. In this model the protein concentrate yield has increased from 28% to 32% of the harvest weight, and about a third of the lipids could now be equivalent to fish oil[22].

Based on algae fermentation literature[23] [24] [25], I have assumed that 60% of the fermentable sugars in the DPJ could be converted to algal biomass, with 50% protein, and 35% lipid content that is high in long-chain omega-3 lipids.

If mixed with the leaf protein concentrate, the combined product would be about 32% of the crop DM.

The algae biomass produced could be about 4% of the crop on a DM basis[26]. The total aquaculture feed produced from a 15tDM/ha crop would be 4.8 tonnes DM, with 50% protein content and 13% fat, including 4% “fish oil” from the algae.


Using the fibre

The fibre fraction is the largest output of the process, produced as a densely compressed mat of fine fibres with about 40% DM content. Our material contained typically 22% cellulose, 21% hemicellulose, 12% lignin, 3% soluble hexoses, 2% nitrogen and 4% minerals.

One thing that could be done with the fibre is to use it as a bioenergy source, by digesting or burning it. The fibre from a tonne of crop DM could produce 7000MJ in a moist-feed burner.

The most energy-intensive part of the high-yield process is the milling stage. In our 87 un-optimised high-extraction runs we found that it consumes typically about 700 MJ per tonne of crop DM. 700MJ is only 10% of the energy available from the fibre, so as a biofuel source the fibre could produce more than enough energy to run the mill and the rest of the process plant as well, perhaps with energy to spare to cover the harvest operation and have a surplus for sale[27].

Alternatively, we could use the fibre as a fermentation feedstock, by hydrolysing it to sugars. Compared with the cellulosic materials commonly used in biofuels developments, our fibre would have some advantages: It would be already present at the site, so accumulation and delivery costs would be already accounted for; and the fibres are finely divided and naturally low in lignin, without any further preparation, so the reaction rates should be rapid and the process equipment relatively small.

If 90% of the cellulose and hemicellulose was hydrolysed to sugars and added to the DPJ[28], and all the microbial biomass produced was mixed with the leaf protein concentrate, the total feed available for aquaculture would be 42% of the crop, on a DM basis[29] [30].

That assumes that the pentose sugars from the hemicelluloses can be usefully fermented, but pentose-based fermentation does exist[31].

If we assume that the microbial oil is equivalent to fish oil[32], the equivalent fish-oil content of the mixed feed would be 14%, which would be higher than found in advanced salmon diets in today[33].


Fish compared with milk powder (updated July 2019)

If we harvested 15 tonnesDM of leaf crop from 1 hectare, we could potentially produce 6.3 tonnes (dry weight) of aquaculture feed with 50% protein content and 20% lipids, with 2/3 of the lipids being microbial oil, rich in long-chain omega-3 fatty acids.

If we fed it to fish and assume a feed conversion ratio of 1.5, we would get 4 tonnes of whole fish.

The same good-quality land used for dairying might produce about 2.5 tonnes of dried whole-milk powder containing 750kg of protein.  (a quarter of the quantity in the leaf protein concentrate)

The 2019 international commodity price of dry whole-milk powder was about 3,000 Euros per tonne, and the price of whole wet Sea Bream (a species similar to NZ’s snapper) was about 5,000 Euros per tonne. At those prices, the export revenue from the fish supported by a hectare of good grassland (20,000 Euros) could be more than double the revenue that the same hectare would produce from whole milk powder (7,500 Euros).

The relative costs of dairy processing compared with aquaculture operation are beyond the scope of this article, but should consider the proportion of value captured by the producer, which should be higher for fish delivered to the consumer or restaurant than it would be for exported milk powder used as an ingredient in the products of multinational companies.

In terms of contribution to global protein supply, the whole fish, after trimming to edible portions, would produce about the same amount of protein that would be obtained in milk from dairy cows on the same land, but it would also contain long-chain omega-3 lipids.


Leaf proteins for food – chloroplastic and cytoplasmic protein

The proteins in leaves can be regarded as being of two classes, originating in the chloroplasts or the cytoplasm of the leaf cell. With some ingenuity they can be separated during the extraction process because they precipitate from the juice at different temperatures. 

The cytoplasmic protein lacks the colour and taste of the chloroplastic fraction, and has higher digestibility. In a trial with Tilapia the cytoplasmic fraction performed as well as fishmeal when replacing 35% of it[34]. It could fetch a premium over the whole protein concentrate, in fish feed formulations that need high protein quality and concentration.

Whole and cytoplasmic leaf proteins are already being promoted and used on a small scale for human consumption[35] [36], which might eventually become popular. The green chloroplastic protein could be used for aquaculture, and the cytoplasmic protein could be used for direct human consumption, perhaps in the way that seed proteins can be extruded into a product resembling surimi or chicken[37].  The quantity of cytoplasmic protein obtained would be similar to the quantity of milk protein obtained from the same land area, but we could get fish as well.


Enabling the use of a wider variety of plants

The protein extraction process also offers opportunities for detoxifying the products of plants that might be plentiful but difficult to use directly in aquaculture. For example, cassava is one of the most commonly grown crops in the world. It is a prolific producer of starch and the leaves have a high protein content. The cyanide it can produce discourages pests and makes it tolerant to some infertile soils. There are ways of removing the cyanide or preventing its production, and the removal process could be conducted efficiently as part of the leaf processing operation.


Economics, resilience and sustainability

The production of microbial oils by fermentation and the hydrolysis of cellulosic biomass to fermentable sugars are still in the early stages of commercialization by various companies.  The leaf-protein extraction operations in France, the only current large-scale producer, appear to be marginally economic. According to reports production volume has been declining, perhaps a consequence of the primary process operation being fuel-intensive dehydration of most of the incoming crop, along with loss of EU subsidies.

In the longer term it is likely that global changes will make leaf crops and the associated biotechnology more attractive.  The high yield relative to seed crops will make better use of what land is available, and allow use of more marginal land.

Resilience in the face of drought, floods and pests will come from the ability of perennials to tap into deep soil stores of moisture and nutrients, and from the ability of leaves to regenerate where seed production would fail. The carbon and nitrogen footprints should be much smaller than those of arable or pastoral agriculture.

The wide diversity of leafy plant species that can be used will improve the reliability of supply, by allowing rapid substitution if one species collapses from a disease or pest incursion.

There is an untapped opportunity for selection of leaf crops for high levels of specific nutrients and flavours.

Because leaf crops can grow successfully in almost any climate, protein production for feed or food could be more local and less subject to global supply and price fluctuations.



[1] Glover JD et al 2010. Increased Food and Ecosystem Security via Perennial Grains. Science 25 June 2010:

Vol. 328 no. 5986 pp. 1638-1639  DOI: 10.1126/science.1188761

[2] http://www.vitalfa.com/Vitalfa_10-21-10_docs/PXaqua.html   Composition and fish performance data from that site is no longer online, but information can be obtained from info at sustainableprotein.com. Some small-scale trial data on fish performance is available at CHATZIFOTIS, S., Vaz JUAN, I., KYRIAZI, P., DIVANACH, P. and PAVLIDIS, M. (2011), Dietary carotenoids and skin melanin content influence the coloration of farmed red porgy (Pagrus pagrus). Aquaculture Nutrition, 17: e90–e100. doi:10.1111/j.1365-2095.2009.00738.x  (http://onlinelibrary.wiley.com/doi/10.1111/j.1365-2095.2009.00738.x/full)

[3] Kaushik S, Aquamax programme director, 2010 (pers. comm.)

[6] Vaughan SR, McDonald RM,  A feasibility study of the production of ethanol by hydrolysis and fermentation of protein extracted lucerne fibre, MAFTech Liquid Fuels Trust Board contract 310/13/1, October 1987 (850 pages). From NZ this data, mean protein extraction over 89 runs was 69.1±4.8%, mean protein concentrate yield was 29% of infeed DM.

[7] Soybean yield and composition 2400kg/ha, 40%CP, 0% NPN. http://onlinelibrary.wiley.com/doi/10.1111/j.1469-8137.1982.tb03303.x/pdf

[8] Example calculation:

Assume lucerne crop 15tDM/ha at 25% crude protein, extracting 70% of the protein and recovering 80% of that as a protein concentrate, giving 2.1 tonnes of protein. Assume soybean typical yield of 2.4 tonnes per hectare at 40% protein giving 0.96 t protein.

[9] Peter D. Nichols, James Petrie and Surinder Singh 2010. Long-Chain Omega-3 Oils–An Update on Sustainable Sources. http://www.mdpi.com/2072-6643/2/6/572/htm

[16] G. N. Festenstein 1972. Water-soluble carbohydrates in extracts from large-scale preparation of leaf protein. Journal of the Science of Food and Agriculture, Volume 23, Issue 12, pages 1409–1415

[17] Peter J. le B. Williams and Lieve M. L. Laurens 2010. Microalgae as biodiesel & biomass feedstocks: Review & analysis of the biochemistry, energetics &economics. Energy Environ. Sci., 2010, 3, 554–590. http://www.eng.utah.edu/~whitty/chen6553/microalgae_biofuel_review.pdf

[18] T.J.Lundquist, I.C.Woertz, N.W.T.Quinn, and J.R.Benemann 2010.  A Realistic Technology and Engineering Assessment Of Algae Biofuel Production. http://digitalcommons.calpoly.edu/cgi/viewcontent.cgi?article=1189&context=cenv_fac

[19] Valeria Cavazzoni, Matilde Manzoni and Renato Craveri, 1988.  Ammonium lactate from deproteinized alfalfa juice byStreptococcus faecium. Journal of Industrial Microbiology & Biotechnology Volume 3, Number 6  p373-376

[20] O. Paredes-López and E. Camargo, 1973.  The use of alfalfa residual juice for production of single-cell protein. Cellular and Molecular Life Sciences Volume 29, Number 10, 1233-1234

[21] Mudgett, R.E.; Rajagopalan, K.; Rosenau, J.R. 1980. Single cell protein recovery from alfalfa process wastes. Trans. ASAE Vol. 23 No. 6  1590-1595

[22] Calculation:

Per 100kg of crop DM: Algae biomass 2.7kg, containing 2.7*0.35=0.95kg lipid. Lipid in total(algae+LPC) = 0.95/27.7 =3.4%

[23] Ratledge C and Cohen Z 2008. Microbial and algal oils: Do they have a future for

biodiesel or as commodity oils? Lipid Technology Vol. 20, No. 7 155-160

[24] Williams and Laurens 2010. http://www.eng.utah.edu/~whitty/chen6553/microalgae_biofuel_review.pdf

[25] Xiufeng Li, Han Xu, Qingyu Wu  2007. Large-Scale Biodiesel Production From Microalga Chlorella protothecoides Through Heterotrophic Cultivation in Bioreactors. Biotechnology and Bioengineering, Vol. 98, No. 4, 764-771. http://onlinelibrary.wiley.com/doi/10.1002/bit.21489/pdf

[26] Calculation:

100kg crop DM produces 25kg LPC. DPJ is 15kg, containing 15*0.3=4.5kg sugars. At 60% conversion, algal biomass = 4.5*0.6 = 2.7kg, containing 1.35kg protein, 0.95kg lipids. Mixed with the LPC the algal biomass would be 2.7/(25+2.7) = 10%

[27] Calculation:

The most energy-intensive part of the plant operation, the pulping and milling operation, is likely to consume 500-800 MJ per tonne of crop DM based on our non-optimised pilot trials. My estimate of total factory energy consumption excluding harvesting and transport is about 2,000 MJ/t crop DM. The energy available from burning the moist fibre (without vapour heat recovery) is about 7,000 MJ/t crop DM[27], which might be enough to also cover the harvest operation and still have surplus energy for sale.

[29] Calculation:

From 100 kg of crop DM get 25 kg LPC containing 12.5kg protein and 2.5 kg lipids. By hydrolysing and fermenting fibre and DPJ get 17kg of microbial biomass containing 8.5kg protein and 6.0kg lipid. Mixture (42kg) contains 50% protein and 20% total lipids. Microbial lipid is 6/(25+17) = 14%

[30] Calculation:

Per 100kgDM: Fibre plus DPJ sugars = 60*0.4 + 15*0.3 = 28 kg sugars. Converted 60% to algal biomass = 17kg protein/lipid complex. This is likely an underestimate: DPJ sugar could be as high as 0.5

[31] Zhao et al 2008. Medium optimization for lipid production through co-fermentation of glucose and xylose by the oleaginous yeast Lipomyces starkeyi. Eur. J. Lipid Sci. Technol. 2008, 110, 405–412

[32] Adam M. Bernstein, Eric L. Ding, Walter C. Willett, and Eric B. Rimm

A Meta-Analysis Shows That Docosahexaenoic Acid from Algal Oil Reduces Serum Triglycerides and Increases HDL-Cholesterol and LDL-Cholesterol in Persons without Coronary Heart Disease

J. Nutr. 2012 142: 1 99-104; first published online November 23, 2011. doi:10.3945/jn.111.148973

[34] Olvera-Novoa, M.A., Campos G., S., Sabido G., M. and Martinez Palacios, CA., 1990. The use of alfalfa leaf protein concentrates as a protein source in diets for tilapia (Oreochromis mossambicus). Aquaculture, 90: 291-302.