Problem - The question, whether it would make sense to reduce meat consumption was posed even thousands of years ago by e.g. Plato and Pythagoras, and still today it is hotly debated by all sorts of scientists. It ought to take into account environmental, ethical, health and basic consumer perception considerations, for instance the environmental pressure of substitutable protein foods; animal welfare thoughts, health aspects including protein contents and valuation, and an often heard statement that (over)consumption of meat in industrialised states causes hunger in developing countries; and finally consumer perception which may include all previous and more straightforward aspects such as flavour and structure.

Objective - It is attempted to find more insight in the environmental impacts of meat and substitutes (protein sources) and shed light on the implications for welfare / environmental optimisation assessments.

Methods - This report focuses on the environmental aspects, having opted to compare Dutch conventionally produced pork, and tofu as a meat substitute. Tofu is no so-called Novel Protein Food (modern meat substitute resembling meat taste and structure to a great extent), though for it is widely used and many data are available, it was chosen. The method used is possibly the most complete for assessing the environmental pressure of a product: life cycle assessment. Keeping in mind substitutability aspects, one need not reflect extensively to conclude that these are no proper replacements, though they serve as an example serving the function of "protein source". Therefore, the unit of comparison is "100 kg of standardised protein". Many previous analyses merely compared protein mass, though it is generally known meat protein is more valuable for man. Hence, a correction using the PDCAAS method was applied - among a variety of methods, this is used by FAO/WHO for protein valuation. For the environmental impact comparison, use was made of previously conducted LCAs. These were somewhat imprecisely, incompletely documented and lacked certain environmental themes such as spatial use and water consumption. Especially these are often debated. It was therefore attempted to shed more light on these aspects, using the previous LCA data, additional (more recent) figures and conversion modules. The LCA was conducted using a self-made model in Excel, which simulates pork and tofu production, i.e. the consumption phase is not included. According to previous research, this phase does not differ significantly for meat and NPFs concerning environmental impacts. In the model, assumptions about e.g. the origin of fodder, pesticide and nutrient use can all be varied by the user, i.e. the model is semi-fixed. Since obviously, the examples do not reflect the image for all meats and NPFs, for a generalisation it is advised to use pre-fabricated LCA software, since creating a model for various substitutes would probably be immensely time-consuming.

Results - The model predicts tofu to be more environmentally friendly in all scenarios, up to a factor 10 per environmental indicator. There is an exception in the scenario where fodder crops are assumed to be entirely non-irrigated: tofu production would use a slightly higher amount of water.

Conclusions - The conclusion, that the meat alternative is more environmentally friendly, becomes apparent when keeping in mind the basic ecosystem structure and functionality: A consumer (animal, higher in the food chain) requires energy, producers fix energy in biomass. Hence, non-equilibrium (Dutch bio-industry here) is clearly less efficient. Linked to this are the bioconversion rates: consuming 1,0 kg standardised protein, requires 6,4 kg pork (41,1 kg pig fodder necessary, containing at least 1,7 times as much standardised protein), or 12,5 kg tofu, which was produced using merely 3,4 kg soybeans (high water content). This immediately points to a practical consideration: it seems environmentally beneficial to consume the fodder instead of meat, though it is clearly non-practical to ingest the fodder mass containing the same amount of standardised protein. The fodder should thus somehow be converted to a more consumable type of food, creating additional environmental pressure.

Concerning the "social scientific" part of the analysis, the main link between LCA and environmental economic theory is established by the allocation question, i.e. the attribution of a bundle of environmental impacts to the variety of products and by-products in a production process. To decide whether allocation on the basis of relative economic value is preferred to different methods such as mass share allocation remains a complex issue. The first seems preferable.

Discussion and recommendations - The meat alternative is clearly more environmentally friendly, though especially for meats in general, more research should be conducted. For instance:

• Cattle digests grass which is non-edible for humans and hence, the conversion rate of fodder:meat is much more favourable in those cases where cows could thrive while it would be impossible to cultivate protein crops for human consumption (e.g. so-called wastelands).
• Animals fulfil a variety of functions (e.g. draught power, companionship, leather, milk, etc.). Utility loss and substitution effects due to a possibly declining animal use should be taken into account accordingly.

Hence, to take into account the wide bundle of product perception aspects (named under "Problem"), it is advised to use multi-criteria analysis. If it could be proven meat somehow causes hunger in developing parts of the world, or that (over)consumption of meat causes great-scale disease-related costs in developed countries (clearly highly complex questions but statements made by several scientists), it seems ethically just to attach high weights to "health" aspects and spend less attention to aspects such as taste - a characteristic possibly valued highly by certain consumers.

Concerning the model and LCA theory, the main discussion points are:

• Certain environmental themes (indicators) are constituted using a partial causal amount of emissions or extractions, due to absent conversion figures or missing data.
• What would be the best method to allocate environmental impacts?
• Some assumptions are still based on constructed data instead of empirical observations, e.g. concerning nutrient assumptions.
• The functional unit (standardised protein) may be changed in order to create a more feasible type / mass of product (see under "Conclusions"), or a different protein valuation method than PDCAAS may be used. The latter is more a question to be dealt with by nutritionists.
• Environmental indicators may be constituted by equivocal units. For instance, it seems more reasonable to find the toxicological effects caused by pesticide emissions "per kg standardised protein per hectare" instead of the usually chosen "per kg standardised protein", since pesticides are diffused over a great surface. Hence, the indicator units as generally found in common LCA literature, should be justified well.

In spite of the results, it cannot be concluded that meat is always or ipso facto less environmentally friendly. It may be more interesting to focus on human needs ("type of consumption system") and on the other, ecosystems ("supply", type of production system). For instance, It seems organic farming links to the carrying capacity better due to the inclusion of environmental, animal welfare and health considerations. Intensive (conventional) systems may disguise substantial external effects.