Background/Question/Methods Deep-soil oak savannas occurred in glaciated landscapes throughout the upper Midwest and high quality remnants are rare due to conversion to human uses such as agriculture and to fire suppression.  The development of protocols for restoring deep-soil savannas have been hampered by the rarity and low quality of remnant tracts.  Prescribed burning has often been employed to reduce woody plant density, canopy cover, and to increase savanna groundlayer alpha diversity.  Few researchers have examined savanna species diversity across increasing spatial scales to determine how patterns of diversity are influenced by fire regime.  We sought to examine the questions of how species-area relationships and savanna indicator species abundance varied across fire frequency and topographic slope gradients in a formerly fire suppressed mesic oak savanna.  We also wanted to confirm that there was evidence of past fire suppression in the contemporary woody vegetation.  Witness tree data from 1834 was utilized to estimate presettlement tree density.  We sampled woody vegetation in 21 0.02 ha plots to determine the influence of fire reintroduction since 1997 (three fire frequency regimes: no fire, low and high) and topographic slopes (flatwoods and slopes) on woody structure and composition.  We also determined present groundlayer vegetation richness and diversity in nested circular plots centered in the middle of the woody vegetation plots in areas 0.25, 1, 4, 16, and 64 m².

Results/Conclusions Past fire suppression was indicated by a more than quadrupling of canopy tree density since 1834.  Seventy-nine percent of savanna indicator species were limited or most important on slopes compared to flatwoods.  Alpha diversity was significantly correlated with fire frequency at all spatial scales and was most highly correlated at the 0.25 m² plot size, whereas it was only significantly correlated with topographic slope at the 64 m² plot size.  Beta diversity decreased with fire frequency in flatwoods but not on slopes.  Richness in 0.25 m² plots explained < 50% of the variation in richness in 64 m² plots.  The ability of a small sampling plot to predict larger scale richness is modest and will be affected by the fire history and topographic position of the plot.  The increase in ground-layer diversity with fire will be greatest at small sales, but will depend on topographic slope and the influence of past fire suppression on the local species pool.  Restoration of flatwoods, but not slopes, will require the reintroduction of savanna indicator species.