Land Use Land Cover changes and their drivers since 1860 within the French northern Alps

Studying Land Use and Land Cover Changes (LULCC) and their drivers is essential for understanding the past origins of current landscapes and anticipating their future evolution. Moreover, such analyses can help prioritize and plan conservation and restoration strategies.

From the Middle Age to the 19^th century in Western Europe, humans - driven by population growth – have altered the natural vegetation succession, notably through forest clearing to establish cultivated fields and pastoral farming systems. Thanks to the industrial revolution and the resulting technical advances which resulted in agriculture mechanization and intensification, important phases of rural exodus happened, inducing waves of land abandonment used for agriculture or pastoralism. Mountainous areas were particularly affected, especially due to their difficult access and their low productivity. The aims of this study were (a) to characterize land-use transitions, with a particular focus on forests, and (b) to analyse the biophysical and socioeconomic drivers of these changes.

We focused on the northern part of the French Alps, which is divided into three eco-regions covering an area of 726 953 ha: Northern Prealps, Inner Northern Alps, External Northern Alps. We used historical maps and aerial photography to analyse LULCC across four periods from 1860 to 2023. Based on the literature, we selected a set of drivers including biophysical drivers (precipitations, temperature, topography, substrate type, avalanches), landscape configuration drivers (distance and percent cover of pre-existing forest, distance to pastoral units, distance to river) and socio-economic drivers (population density and change, distance to settlement, road density, tourism density, number and rate of change of pastoral units). This dataset allows us to analyse land-use changes over a long-time span (approximately 160 years) and at a large spatial scale.

To analyses changes between different dates, we used a grid of systematic points, with a density of one point per hectare, to generate several transition matrices. To assess the effect of drivers on LULCC we performed logistic regression models. Specifically, we fitted models of LULCC or forest recovery as a smooth or linear function of the different drivers. We also took into account potential biases in the results of the different models related to spatial autocorrelation of observations by integrating distance based on the sample variogram of the residuals obtained from a model without spatial dependence. Overall, we expect a global forest expansion, with positive effects of lower precipitations, steeper slopes, substrate type (hard), lower population density and pastoral decline, and negative effect of high tourism density and avalanches. Preliminary results will be presented at the conference.