Disturbance and environmental change may cause communities to converge on a steady state, diverge towards multiple alternative states, or remain in long-term transience. Yet, empirical investigations of successional trajectories are rare, especially in systems experiencing multiple concurrent anthropogenic drivers of change. We examined succession in old field grassland communities subjected to disturbance and nitrogen fertilization using data from a long-term (22-year) experiment. Regardless of initial disturbance, after a decade communities converged on steady states largely determined by resource availability, where species turnover declined as communities approached dynamic equilibria. Species favored by the disturbance were those that eventually came to dominate the highly fertilized plots. Furthermore, disturbance made successional pathways more direct under low nutrients, revealing an important interaction effect between nutrients and disturbance as drivers of community change. Our results underscore the dynamical nature of grassland and old field succession, demonstrating how community properties such as beta-diversity change through transient and equilibrium states.

Data were collected in successional grassland fields at the Cedar Creek Ecosystem Science Reserve in Minnesota, USA (CDR, Lat: 45.4 Long: 93.2 W) from 1982 to 2004. In 1982, identical disturbance X nutrient addition experiments were established in three abandoned agricultural fields within 5 km of one another that were last tilled and farmed in 1968 (Field A), 1957 (Field B), and 1934 (Field C).

Within each of the three Fields, two grids (35 X 55 m) were established in 1982 for nutrient application, one in an area that was thoroughly disked in the spring of 1982 (E002), and another in an adjacent area that remained intact with old field vegetation (E001). The disking treatment pulverized the existing vegetation, leaving bare soil which was then raked to remove clumps of vegetation. Each grid consisted of 54, 4 X 4 m vegetation plots, receiving one of nine nutrient treatments (applied annually in May or June) in a randomized block design, with 6 replicate plots per field. A nearby remnant grassland within a native oak savannah (Field D) that had never been clear-cut or plowed was also surveyed annually and provides a comparison for our study.

The nutrient-addition treatment had nine levels representing different combinations of Nitrogen (0 – 27.2 g N yr-1 added as NH₄NO₃) and Other Nutrients (20 g m-2 yr-1 P205; 20 g m-2 yr-1 K20; 40 g m-2 yr-1 CaCO₃; 30.0 g m-2 yr-1 MgSO₄; 18 ug m-2 yr-1 CuSO₄; 37.7 ug m-2 yr-1 ZnSO₄; 15.3 ug m-2 yr-1 CoCO₂; 322 ug m-2 yr-1 MnCl₂; and 15.1 ug m-2 yr-1 NaMoO₄).

Beginning in 1982, vegetation was sampled by clipping a 10 X 300 cm strip each year within each plot at the ground level. After clipping, biomass was sorted into the previous year’s growth (litter), and the current year’s growth (live biomass). Live biomass was sorted by species, dried, and weighed to the nearest 0.01 g.  All plots in all fields were sampled annually with the exception of years 1995 (only E001 sampled), 2001 (only E001 sampled), and 2003 (only E001 and Field C E002 sampled). Due to new treatments (experimental burning and fence removal) among the three experimental fields after 2004, we restricted our analyses to the time period 1982 - 2004. Additionally, beginning in 1992, three randomly chosen replicate plots within each nutrient treatment in the E002 grid received nutrient cessation and experimental burning, which we excluded from our analyses from 1992 onwards. We conducted statistical analyses on plot-year combinations with original disturbance X nutrient treatments established in 1982. Prior to all multivariate analyses, we applied a ln(1 + x) data transformation where x = biomass (in g) of individual plant species within a plot in a given year.

Code for analyses and figures in the manuscript can be found here: https://github.com/melissadesiervo1031/CedarCreekconvergence