Many ecosystems, such as the prairies and savannas in the Midwest region of the United States, require fire as regular disturbance. Fire management is important for maintaining these ecosystems; however, varying burn frequencies can alter the outcomes. We test how burn frequencies affect plant community dynamics, specifically; lost and gained plant species over time.A series of burn units established at Cedar Creek Ecosystem Science Reserve, University of Minnesota have vegetation data on frequencies ranging from never burned to burned every nine in ten years. Randomly spread throughout 364 hectares, plots were surveyed every five years from 1984 to 2010. The plots were divided into frequency categories of high (3/4 to 9/10 years), low (1/10 to 1/4 years), intermediate (1/3 to 2/3 years), and controls. These were analyzed by total species gained and lost as well as changes by functional group (grass, forb, shrub and tree) and endemic ecosystem (prairie, forest, and disturbed site). The data were non-normally distributed, thus Kruskal-Wallis tests were used. Each dependent variable was tested for co-linearity with none determined to be significant within a classification system. Pairwise Comparisons used an alpha of (.017) to account for double classification between the functional groups and endemic ecosystems.
The analyses showed six variables significantly influenced by the time interval: Prairie Loss (P=.014), Forest Gain (P=.001), Forb Gain (P=.002), Grass Gain (P=.014), Shrub Gain (P=.000), and Shrub Loss (P=.003). Nine variables were significantly related to the burn frequency: Prairie Gain (P < .001), Prairie Loss (P=.001), Forest Gain (P=.024), Forest Loss (P=.048), Forb Gain (P=.025), Grass Gain (P=.012), Grass Loss (P=.006), Shrub Gain (P=.027) and Shrub Loss (P=.006).