Long-term effects of haying and prescribed fire on the composition and diversity of wet prairie plant communities

Abstract

Plant species composition and diversity in many terrestrial ecosystems depend on frequent disturbances. Management of these historically disturbance-dependent habitats often requires replicating past disturbance regimes or implementing management approaches that mimic their ecological effects. For example, efforts to manage North American tallgrass prairie frequently utilize prescribed fire to maintain these historically fire-dependent grasslands. However, alternatives to prescribed fire, such as haying, have attracted the interest of conservation practitioners and landowners. The paucity of long-term experiments, especially in wet prairies and sedge meadows, limits our understanding of how these different management techniques influence the composition and diversity of such perennial-dominated plant communities. We conducted a 23-year experiment within a remnant wet prairie complex in northwestern Minnesota, USA, to evaluate the effects of haying and prescribed burning on plant species richness and community composition. Experimental treatments-no active management (control), annual late-summer haying, spring burns at 4-year intervals, and fall burns at 4-year intervals-were applied continuously between 1999 and 2021. Spring burning at 4-year intervals tended to maintain the highest species richness while fall burning at 4-year intervals supported intermediate levels of species richness similar to control plots. For the first 10 years of the experiment, species richness in the annually hayed plots was comparable to the burned and control plots. Thereafter, richness declined rapidly in the hayed plots, leading to a 30% drop in species richness between 2010 and 2021. Annual haying also altered plant composition at a rate that far outpaced compositional changes in other treatments. Temporal shifts in plant community composition reflected idiosyncratic species' responses to experimental treatments. Many of these species-specific responses to treatments were delayed, becoming apparent more than a decade after treatments were initiated. Annual haying and prescribed fire have distinctive long-term ecological effects on wet prairie plant communities and should not be considered interchangeable management options. Continuous annual haying erodes plant diversity in wet prairies and shifts plant composition while periodic spring fires tend to maximize local plant richness. This study exemplifies the importance of conducting long-term experiments to better inform management of disturbance-dependent habitats.

Keywords: brush prairie; disturbance; fire; haying; long‐term experiment; mowing; sedge meadow; tallgrass prairie.

FIGURE 1

(a) Map illustrating the study location, distribution of treatments within (b) northwest and (c) south experimental grids, and (d) within‐plot sampling design. The study took place at Pembina Trail Preserve State Natural Area in northwestern Minnesota, USA. This site is in the Agassiz Beach Ridges landscape at the eastern edge and northern extent of the tallgrass prairie region (historic extent of tallgrass prairie shaded in gray). We established two grids—a northwest grid and a south grid—composed of 10 × 10 m cells spanning two subsets of the wet prairie complex at Pembina Trail Preserve SNA. Experimental treatments were applied to a stratified random sample of 30 × 30 m units. We conducted vegetation sampling in N = 32 units (N = 8 units per treatment). All vegetation sampling was conducted in the center 10 × 10 m grid cell of the experimental unit (vegetation sampling plots outlined in white) to minimize potential edge effects. Within each vegetation sampling plot, we sampled vegetation within 1 × 1 m subplots (N = 20 per plot). The same set of subplots (shaded in black) were used to sample vegetation throughout the full duration of the experiment within each plot. Note that the illustration does not reflect the proximity of the northwest and south grids; they are separated by ~1.5 km.

FIGURE 2

Long‐term changes in total plant species richness per plot during the 23‐year experiment. Annual haying begins to erode species richness after 10 years of experimental treatment. Thick solid lines and shaded bands with dashed upper and lower bounds depict predicted species richness and 95% CIs, respectively, from the linear mixed‐effects model. Thin lines depict observed species richness within each study plot.

FIGURE 3

Nonmetric multidimensional scaling (NMDS) ordination depicting changes in plant species composition within plots and among experimental treatments over the 23‐year experiment. Thick solid vectors depict the start and end composition of each study plot in ordination space. Thin vectors depict the trajectory of each study plot during the experiment.

FIGURE 4

Rates of temporal plant community turnover during the 23‐year experiment. We quantified the plot‐level dissimilarity between initial community composition at the start of the experiment and each subsequent vegetation survey. Annually hayed plots diverged more rapidly from initial species composition than other experimental treatments. Thick lines depict predicted similarity from a linear mixed‐effects model while thin lines depict observed similarity values within each study plot.

FIGURE 5

Species‐specific responses to experimental treatments over the 23‐year experiment. For the 24 most abundant taxa, we plot the observed and predicted changes in plot‐level species frequency. Thick solid lines depict frequency values predicted by the multilevel model, while thin lines depict observed frequency within each study plot. See Appendix S1: Figure S3 for graphical depiction of all species included in the multilevel model (MLM).