A 6-year experiment examined the effects of spring and summer fires on grasses in southern Wisconsin. Synthetic communities of C3 and C4 grasses were seeded (100 seeds m -2 species -1 ) in 1992 and subjected to prescribed burns in May and August of 1995 and 1997, or left unburned. By 1994 all plots were virtual monocultures of the C 3 reed canary grass (Phalaris arundinacea L.). By the second post-season sample in 1998, total productivity of plots burned in May was higher (781 ± 212 se g m -2 y

more »

... e a r -1 ) than those burned in August (362 ± 28 g m -2 y e a r -1 ) or left unburned (262 ± 43 g m -2 year -1 ) due to the incursions of either the C 4 grasses big bluestem (A n d r o p o g o n g e r a r d i i V i t m a n ) , switichgrass (Panicum virgatum L), or both. These large late-season grasses are much more productive per area covered than P. arundinacea or the other two C 3 grasses present, Elymus virginicus L. and Poa pratensis L. Even at this early stage of succession, C 4 production in plots burned in May was 5 to 6 times that in the other 2 treatments. August burns produced a mix of C 3 and C 4 grasses but did not strongly favor the pre-treatment C 3 dominant P. arundinacea. Unburned plots most resembled those burned in August in species composition, but differed in having 4 times the accumulated litter, perhaps foretelling divergence in C 3 and C 4 composition as succession proceeds.