Grasslands, by definition, are dominated by graminoids.  Nevertheless, forbs also make up a substantial part of vascular plant diversity in grasslands and are important resources of mammalian herbivores. However, forb recruitment is constrained by successful dominant graminoids, limiting access to safe sites for germination. Disturbances created by herbivores can reduce graminoid dominance and favor forb recruitment. Here we hypothesize that intense disturbance, such as that caused by megaherbivores, promotes safe sites for forbs in such graminoid-dominated grasslands, whereas disturbance by today’s herbivores, such as small rodents, may not be sufficiently intense. We selected a total of 80 plots with either of four successful graminoid species in tundra grasslands of the Varanger Peninsula, Norway. The graminoid species were silicon-poor or rich, and of either mat- or bunch growth form. Plots were further selected in both rodent disturbed and undisturbed areas. We manually removed the dominant graminoid in half of the plots, mimicking megaherbivore disturbance by reducing both shading capabilities and belowground rhizome and root systems. Results show that forb recruitment was significantly enhanced one year following the manual removal of all four graminoids. This effect on forb recruitment was similar among the four graminoids even though they were associated with distinct plant communities. The rodent disturbance did not enhance forb recruitment. In plots with rodent disturbed graminoids, the manual removal enhanced forb recruitment only in plots with silicon-rich graminoids. Forb recruitment was further enhanced by higher levels of initial species richness, initial forb abundance and soil moisture. Our findings support the hypothesis that intense disturbance, simulating megaherbivore effects on dominant graminoids, significantly enhances forb recruitment.

This study was conducted in riparian grasslands along tributaries of the Bergebyelva river, Varanger Peninsula, Norway (70.30^o N, 29.02^o E). 

The main treatment was manual disturbance of dominant graminoids that made up most of the vascular plant abundance in grassland vegetation plots. Both aboveground and belowground plant parts were removed, and in doing so most other vascular plants present in the plots were also disturbed. To test the extents to which small rodent disturbance, dominant graminoid growth form and dominant graminoid silica content affected the outcome of this manual disturbance, we established a nested study design in two riparian valleys (Torvhaugdalen, 70.31o N, 29.08o E and Bearalveaijohka 70.29o N, 28.96o E) in July 2023. In both valleys, we identified substantial populations of the four dominant graminoids: the silica-rich bunch graminoid Deschampsia cespitosa, the silica-poor bunch graminoid Carex nigra, the silica-rich mat graminoid Calamagrostis spp. and the silica-poor mat graminoid Anthoxanthum nipponicum. For each dominant graminoid within each site, we identified areas with no visual signs of rodent disturbance and areas with high levels of rodent disturbance (hereafter denoted as Non-rodent and Rodent Disturbed). Disturbance was identified  by vegetation clipping, runways, and feces presence. In each area we established a minimum of two 0.5m x 0.5m plots, or plot pairs. The plots representing a single pair were at least 0.5m from each other and plot pairs that were either Non-rodent or Rodent Disturbed were at least 7m from each other. 

Initial plot conditions in July 2023 were determined by recording a list of all vascular plant species as a measure of species richness, estimating the percent cover per species, measuring plant height (in each plot quadrant), and assessing rodent disturbance intensity–percentage of the plot with graminoid clippings or runways and feces presence. After these initial recordings, we randomly chose one plot per pair for the manual disturbance. To simulate megaherbivory capable of disturbing both above- and belowground plant structures, we removed both above- and belowground biomass – breaking shade aboveground and rhizome and belowground root and rhizome networks (hereafter denoted Manual). 

In July 2024, we revisited each plot, assessed percent cover per species and plant height as in 2023, and counted the number of seedlings in the plot (many seedlings had developed their first true leaf, allowing us to distinguish between forbs and other Dicotyledoneae plants - e.g., Salix shrubs). We did not register graminoid seedlings. This was because we could not differentiate clearly between true seedlings and regeneration from buds. Indeed, there were likely remains of rhizomes in the soil after the manual disturbance that have likely survived the winter. 

We also sampled plot level abiotic information. We measured photosynthetic active radiation (PAR) at seedling height and above the vegetation canopy using the quantum sensor on a LiCOR 600 Porometer/Fluorometer (LiCOR Biosciences, Lincoln, Nebraska, USA) to determine the relative amount of light reaching seedlings. The PAR index ranged from 0 to 1, 1 representing  a situation where all light reaches seedling. We further measured percent soil moisture using a SM150 soil moisture sensor (Delta-T Devices, Cambridge, UK), taking the mean of three measures per plot. 

We calculated species richness and Simpson’s (1949) diversity index for each plot for initial conditions in 2023 before any manual treatment. We then estimated indices for species turnover in relation to species identity and dominance shifts, considering initial plot conditions in 2023 and change observed in 2024. We used the species exchange ratio based on richness (SERr) and abundance (SERa) per plot (Hillebrand et al. 2018) to evaluate species turnover per plot.