Biodiversity footprints of 151 popular dishes from around the world

recipes were broken down into their raw ingredients sourcing additional recipes (S6 Table). One-way ANOVA was used to compare the biodiversity footprint between different categories of dishes (vegan, vegetarian and dishes containing meat). Tukey tests were used to correct for multiple comparisons. Scenarios The biodiversity footprint of all dishes was assessed in both locally and globally produced sce- narios. In the locally produced scenario, biodiversity footprint data from the selected country the dish was typical of was used. For the globally produced scenario, the global distribution of the crop was used to obtain a global biodiversity footprint without considering countries’ borders. Dishes containing meat from livestock could have been produced using extensive methods based on pasture feeding or using intensive feedlot methods. We thus considered these meth- ods as separate scenarios (pasture-grown and feedlot-grown scenarios) using the dataset that considers biodiversity footprint per unit of area of pasture and biodiversity footprint per ton of crop produced respectively [26]. The ingredients’ biodiversity footprint per kcal per gram under all three biodiversity indicators were also calculated. For the feedlot-grown livestock, the biodiversity footprint of the livestock was calculated through the corresponding weights of crops constituting the livestock’s feed (S7 Table). The crop weights were obtained after multiplying the conversion ratio of the respective feeds with the weight of the livestock ingredient. The conversion ratio is the ratio of feed weight con-

sumed to weight of livestock ingredient produced. After substitution of the livestock ingredi- ent with its respective feeds, the biodiversity footprint of the livestock ingredients was calculated as the sum of the biodiversity footprints of the crops. Under the pasture-grown sce- nario, the biodiversity footprint of a livestock ingredient was calculated as: Biodiversity footprint of a livestock ingredient ¼ Lm � Fm � Wm Cm ; ð2Þ where Lm is the biodiversity footprint of the pasture to feed livestock m (in number of species ha-1, number of threatened species ha-1 or range rarity ha-1 respectively for each metric consid- ered); Fm is the pasture area (in ha) required to produce one head of m (S8 Table); Wm is the weight of the livestock ingredient required in a given dish; and Cm is the carcass weight of one head of the livestock respectively (S9 Table). Pasture areas required to produce a head of live- stock were based on countries or regions where the respective livestock production was the highest (S8 Table). The rationale for this assumption is that countries with large production pasture-fed livestock will present professional production systems in which the utilized area would be optimized, being a conservative estimate. Ambiguous ingredient types and measurements When an ingredient was specified in a generic manner, the top ingredient of that type in terms of global production was assumed. For instance, soybean oil was assumed when only “oil” was mentioned in the recipe. White sugar was chosen for unspecified sugar type. To produce

sugar, sugar beet was assumed as the sugar crop for all countries except tropical countries where sugar cane was assumed. Black pepper was selected as the pepper type and milk was assumed to be cow’s milk. Standardisations of the recipes’ measurements were also made when weights were not spec- ified. In recipes which stated “a pinch” as a form of measurement, they were standardised to be 0.5 g. Similarly, the amount of oil used for stir-frying was taken to be 1 tablespoon ~14 g per PLOS ONE Biodiversity footprint of popular dishes PLOS ONE | https://doi.org/10.1371/journal.pone.0296492 February 21, 2024 5 / 15