Fig. 1

The methodology can be outlined through a flow-chart which describes the main logical steps followed in the discussion. At the beginning (a) our sample was divided in two subsamples, one formed by the farms with agritourism and the other by the farms without agritourism. On the second hand, the data analysis was performed both on national sample and on sample sections according to the altitude of the farms. The first sample matches the ones (Y = 1) in our dependent variable and the other matches the zeros (Y = 0). The dichotomous structure of the dependent variable requires a binomial logit model (b) to be associated to the matrix of explicative or independent variables. This matrix was built by selecting a set of environmental indicators as reported in detail in Additional file 1: Table S1. At last (c) the tests applied to assess the fitness of the model may give good or bad results, according to the more or less explicative power of the variables included in the matrix, therefore some of them may be excluded or not from the model