Erosion rate response to the reduction of forest cover following metallurgical activities on Elba Island (Italy)

Latest with the onset of the Neolithic period, humans have modified the landscape by farming, extracting natural resources, and reducing forest cover. Many studies have demonstrated that these human activities can increase the vulnerability of soils to erosion. However, the magnitude and duration to which such activities have reduced forest cover and accelerated soil erosion have been mainly based on proxy data derived from sediment and pollen records. In this study, we test a new methodological approach by coupling leaf wax isotope analyses with erosion rate measurements based on ¹⁰Be and ¹⁴C to investigate the impact of human activities during the Holocene on vegetation and erosion rates on the island of Elba, Italy. We hypothesize that meeting the substantial fuel demands for iron smelting on Elba Island, initiated by the Etruscans and subsequently continued by the Romans from the 4^th century BCE onwards, exerted significant pressure on the island's forest ecosystems. This reduction of forest cover likely accelerated soil erosion processes, driven by the removal of vegetative cover essential for soil stability. To test our hypothesis, we measured cosmogenic ¹⁰Be and ¹⁴C in quartz from stream sediment samples to reconstruct changes in erosion rates during the Holocene. We complemented this with leaf wax isotope analysis (δ²H, δ¹³C) from floodplain sediment cores to explore vegetation and hydrological changes. Due to the shorter half-life of ¹⁴C compared to ¹⁰Be (~5.7 kyr and ~1.4 Myr, respectively), ¹⁴C records erosion on shorter time scales than ¹⁰Be. Apparent erosion rates obtained from the two nuclides show a marked offset: ¹⁴C apparent erosion rates of 129 to 1080 mm kyr ^-1 are up to two orders of magnitude faster than the corresponding ¹⁰Be erosion rates of 20 to 50 mm kyr ^-1. This discrepancy can be explained by a substantial increase in erosion toward modern times. To identify the timing of erosion change and thickness of soil loss, we apply a Markov chain Monte Carlo inversion with several simple erosion histories. These data allow us to compare the history of land use with vegetation change and the erosion response on a landscape scale.