After 32 years of high nitrogen addition at Cedar Creek, plot halves where addition stopped gained 3.3 plant species per strip with annual spring fire and 0.8 species without fire (2015–2022)

Question: does annual spring fire change the recovery of grassland plant richness after a long nitrogen (N) enrichment stops? Approach: in 2014 the Cedar Creek experiment E002 stopped fertilization on one half of each fertilized plot in fields B and C, after 32 years of N addition at 0 to 27.2 g N m-2 yr-1. In field B, three of six plots of each N level have been burned each spring since 1992. We used the public E002 biomass table (revision 14) and compared the two halves of each plot in each sampled year from 2015 to 2022 (408 pairs, 72 plots). The plot was the unit of inference. We tested fire with a randomization test of fire labels within N level. Findings: the within-plot richness gain increased with the previous N rate (0.79 species per strip per unit ln(N+1); 95% interval 0.55 to 1.02). At the three highest N rates in field B, stopped halves gained 3.33 species per strip in burned plots and 0.81 species in unburned plots (difference 2.52; 95% interval 1.81 to 3.26; p = 0.0003). Fire alone did not change richness in halves that stayed fertilized (difference -0.04). With fire, stopped halves closed 48% of the deficit to unamended controls; without fire they closed 17% (field B) and 26% (field C). Native taxa gave 2.46 of the 3.33 species gain, with native C4 grasses giving 0.93. Fire also gave a larger decrease of the introduced grass Agropyron repens and of total biomass, and it removed almost all litter. Plots in field C where N stopped in 1992 still had 3.4 fewer species per strip than controls in 2015-2022. The results held when we removed genus-level taxa, kept woody taxa, removed 2018, or used only years common to both fields, and in Poisson and mixed models. Limitations: one field and one fire regime; 0.3 m2 strips; data end in 2022; the mechanism (litter removal, loss of the dominant grass, or fire benefit to C4 grasses) is not identified; plot coordinates were not available to test spatial dependence.

nitrogen enrichment · cessation · recovery · prescribed fire · grassland · species richness · hysteresis · Cedar Creek LTER · split-plot · randomization test · litter · Elymus repens · C4 grasses · restoration