Tussock size and annual grass productivity
To determine tussock size of P. ligularis and P. speciosa , aboveground volume of the tussocks was calculated using the equation of Derner et al. (2012):
V = 1/3 Πh (r 2 + rR +R 2)
where V is the volume, h is the distance from the soil surface to the top of tussock (height), r is the radius of the basal area occupied by tussock, and R is the radius of the area at top of tussock. We used this equation because the circumference of tussocks increases from the base to the top. Since a high number ofP. humilis tussocks was observed in the +N and +NW plots four years after the start of the experiment, this species was also included in this study. However, in C and +W plots we did not find an adequate number of P. humilis tussocks to consider sampling and further statistical analysis. Determinations were made on randomly selected tussocks in December 2019.
To determine the annual grass productivity, a 50 x 50 cm quadrat was randomly selected in each plot, taking into account that if the quadrat fell on a shrub individual, it was selected again to avoid shrubs in the samples. This is due to the difficulty of estimating annual shrub growth using this method, as there are few shrub individuals per unit area, which would overestimate the calculated ANPP. All aboveground plant biomass was cut at the soil level and transferred to the laboratory, where green biomass, dry biomass, and litter were separated. The plant biomass was then oven-dried at 70 °C for 72 h. Cacti and herbaceous dicotyledons were excluded from the analysis due to their low representativeness in the samples (<1%). Subsequently, biomass was weighted and ANPP was determined as the biomass of each type divided by the quadrat area in each year and expressed as g m-2 yr-1. Productivity was measured once a year during peak green grass biomass (December) from 2013 to 2019. We chose to harvest in this month because both temperature and soil moisture are favorable for growth, while later (summer) the temperature is higher, but soil water availability decreases and atmospheric water demand increases. Moreover, a study applying eddy covariance techniques in the same Patagonian steppe shows the peak of gross primary productivity and ecosystem net carbon exchange in December (Burek et al. 2023, under revision). Sala & Austin (2000) determined that the method for ANPP estimation that introduces the least error in grasslands with marked seasonality is that consisting of a single annual harvest.

Shrub size and annual shrub productivity

Three individuals of each of the most abundant shrub species at the site (A. prolifera , A. volckmannii , and S. filaginoides ) were selected and targeted per plot in 2015. In December of each year from 2015 to 2019, the height and major and perpendicular diameter of each selected individual were measured. To determine the major and perpendicular diameter of each, photographs were taken of the top of the canopy including a scale, and the images were analyzed with the ImageJ program (Ferreira & Rasband, 2012). The biomass of each individual was estimated by applying allometric relationships estimated by Oñatibia et al. (2010) for the same species and in an area close to that of the present study. The equations used were the following:
b = 0.000003 x dh 3.9321 for A. prolifera
b = 0.0002 x dh 3.1402 for A. volckmannii
b = 0.0002 x dh 3.1876 for S. filaginoides
where b is the estimated aboveground biomass (g) and dh is the average diameter between the major and perpendicular plus plant height (cm). Annual growth was determined as the change in estimated biomass between consecutive years and expressed per area. ANPP was estimated as annual growth per area multiplied by the percent cover of each species at the site. When dead individuals were observed among those marked, this was recorded and the number of dead plants per year was determined. In that case, new individuals were marked and measured. If plants considered dead sprouted in subsequent years, they were removed from the dead plant category.