Environmental variables
Temperature and precipitation data were obtained from a meteorological station located 2 km from the study site (http://sipas.inta.gob.ar/). The volumetric soil water content was continuously determined by sensors using the dielectric constant of the surrounding soil medium (ECH2O, Dielectric aquameter, Decagon Devices, Inc). In each plot, sensors were installed at 10, 40, 100 and 200 cm depth. The sensors were connected to CR1000 automatic data acquisition systems (Campbell Scientific Inc.) powered by batteries charged by a solar panel. Data were recorded every 30 min. Soil psychrometers (PST-55, Wescor, Logan, UT) were used to continuously monitor soil water potential at 40 cm depth, which was measured every 30 min and data were recorded with a datalogger (CR-7, Campbell Scientific Inc.).
Data for pH, total nitrogen, inorganic nitrogen, available phosphorus, and carbon content in the upper soil layer were obtained in November 2018. Three soil cores (5 cm diameter, 5 cm depth) were randomly taken from each plot and mixed well to obtain a composite sample (n = 5). The pH was determined in a 1:2.5 solution of distilled and deionized H2O. Soil organic matter, total nitrogen and available phosphorous were analyzed by the Soil Laboratory (INTA Chubut, Argentina). Total nitrogen was determined by the Kjeldahl method (Bremner, 1996), inorganic nitrogen by the distillation method (Bremner & Keeney, 1965), phosphorus by the Olsen method (Olsen et al., 1954), and organic matter by the wet oxidation method (Walkley & Black, 1934). The C content of the organic matter was then determined using the factor proposed by Van Bemmelen, of 1,724, which assumes that 58% of the organic matter is composed of C (Tabatabai, 1996). The C/N ratio was also calculated by dividing the carbon content by the nitrogen content (Bremner, 1996).

Vegetation cover and leaf nitrogen content

Changes in vegetation cover between 2012 (before the start of the experiment) and 2019 (i.e., 6 years after the start of the experiment) were determined using the intercept point method (Mueller‐Dombois & Ellenberg, 1974). For this, the plant interception was determined every one meter along the two diagonals of each plot. Subsequently, the percent cover of each species and bare soil in each plot was calculated annually by dividing the number of positive touches for each species or bare soil by the total number of touches. It should be noted that ephemeral herbaceous species, mainly present during October and November, were not taken into account in the vegetation cover analysis.
Leaves were collected in December 2017 for analyses of nitrogen concentrations. Fully expanded and healthy leaves from different individuals of each species and in each plot were collected to obtain a composite sample per plot and species. Total leaf N content was measured using the Kjeldahl technique.