*2.2. Water Footprint Estimation*

The water footprint (WFP) of a group of consumers can be expressed in terms of water volume per unit of time per capita. For food products, to assess the water footprint we need to consider the process of growing the crop or tree. The total WFP of the process of growing crops or trees is the sum of the green, blue, and grey components. The green (rainwater) and blue (surface and groundwater) components in crop water use are calculated by accumulating the daily evapotranspiration over the total growing period. The grey water part represents the theoretical volume of water required to dilute all the pollutants released during the production in order to achieve a water quality standard [26]. To assess the water footprint of food products of animal origin consumption, we resorted to the water footprint network database, which takes account of the blue, green, and grey water footprint of animal origin products.

The assessment of the consumer's water footprint is based on the methodology developed by the water footprint network described in the water footprint assessment manuals [26,38]. The process begins with an inventory of water requirements for each stage of the product production and processing. The water footprint of a "process step" forms the basis of all water footprint inventories. The water footprint of a good is the aggregate of the water footprints of the various relevant stages in the production of that good. A consumer's water footprint is the sum of the water footprints of the different consumed products. The water footprint of a group of consumers can be expressed in terms of the volume of water per unit of time per capita.

As previously mentioned, to assess the water footprint of the main groups of food products in the different regions of Tunisia, we used food and nutrition survey data published in 2015 as well as the database developed by Mekkonen and Hoekstra [19,39] on the water footprint of crop production as described in detail in [18]. Concerning food products of animal origin, the water footprint includes both the water footprint of feedstuffs and the water directly used for breeding animals and for processing dairy and meat products. A database, grouping the majority of the water footprints of animal products is also available on the Water Footprint Network website [19,39]. However, it is important to note that one of the limits of this work, which affects the precision of the estimates, is the absence of data on the water footprint of fish and seafood, an important component of the Tunisian diet. For more details about determination of the main food products and the water footprint assessment method used, we refer readers to Souissi et al. [18], which presents these steps in detail.

The water footprint may depend on the specific characteristics of a household that usually influence food consumption patterns. The INS survey also includes data on households' socio-economic and demographic characteristics such as region, family size, income, etc. As we noted previously, according to the literature such characteristics influence diets within households [27–29]. The characteristics of the households in the sample are presented in Table 1. The sample includes both urban and rural populations from all regions of the country. We note that more than 62% of the sample live in urban areas. There are eight different income categories, ranging from less than TND 500 to more than TND 4500 (exchange rate during the year of data collection: 1 USD = 1.44 TND). The most common category of employment is laborer. The most common household size is 5–6 members. About 14% of the respondents are classified as poor. The poverty line is estimated using the methodology of the World Bank. The monetary approach is used to determine this poverty line (or minimum income), which will be considered as the absolute poverty standard. The population categories with a level of income below this threshold have a high risk of living in conditions of absolute poverty. This method is based on the assessment of the cost necessary to meet the minimum calorie requirement of a low-income segment of the population. This is the category of the population living with a per capita income hovering around the 20th percentile of household spending. The 20th percentile is the level of per capita income below which the poorest 20% of the population live. The head of household is typically a married man. About 29% of the respondents were illiterate.
