1994 Proceedings
North American Conference on Savannas and Barrens

Savanna and prairie occur in central North America between the eastern deciduous forest and the Rocky Mountains. Predominantly herbaceous, prairie is classified into three types by rainfall and consequent grass composition. The easternmost and moistest division is the tallgrass prairie (Risser et al. 1981). Savanna habitats are diverse in composition and structure, contain a mixture of woody and herbaceous components, but lack a widely accepted definition (Nuzzo 1986). Although tallgrass prairie once broadly covered the middle United States (Fig. 1), it has been >99% destroyed, with savanna similarly destroyed (Curtis 1959; Nuzzo 1986). Patches of original prairie and savanna vegetation remain in preserves, parks, and under utilized or fallow farm and forestry lands. These remnants are usually isolated and often degraded. Because of this catastrophic habitat loss, butterflies (as well as other species) obligate to prairie and savanna are rare, often restricted to preserves. They typically associate with the herbaceous component of these habitats. The Karner Blue (Lycaeides melissa samuelis) is federally endangered and the Dakota Skipper (Hesperia dacotae) and the Regal Fritillary (Speyeria idalia) are federal candidates for listing. Various butterflies are also on state endangered/threatened lists.

Prairie and savanna require certain climatic conditions, soil types, and/or disturbance processes (such as animal herbivory and fire) to exist (Vogl 1974; Anderson 1982). However, these processes are nearly entirely disrupted today because of the destruction and fragmentation of these habitats. In unmanaged prairie and savanna sites, the vegetational composition and structure often alter as a result of invasion and maturation of woody species and smothering under dead plant matter. Thus, these habitats usually require active management today to maintain the ecosystem and its biodiversity. But because of their decimation and degradation, it is difficult to know exactly which processes occurred in which ways to maintain these habitats. Various theories exist to explain the origin and maintenance of prairie and savanna habitats. Despite this conflict, the management most successful at maintaining these habitats and their native species is logically assumed to be based on these natural processes.

STUDY AREA

The study sites include 93 prairies in five states (Fig. 1) diverse in vegetation type, quality, and size. Most are preserves managed principally with fire, with burns averaging about 25% (range 0-99%) of the prairie patch per year. Some Missouri preserves are managed primarily with summer haying and a little burning and cattle grazing. Sites in Illinois, Missouri, and Wisconsin fall within the zones both for tall-grass prairie and oak savanna (Fig. 1). Depending on scale and definition, habitat patches in many sites could be classified either as prairie or savanna. We have also surveyed >80 pine-oak barren sites in central and northwestern Wisconsin within range of the Karner Blue. Variously considered brush prairie, sand barrens, pine barrens, pine-oak openings, and oak savanna and diverse in floristic composition, structure, and management history, these sites occur within state and county forests, wildlife areas, roadside and powerline rights-of-way, army reservations, and national wildlife refuges. A few more Wisconsin barren sites in national forest and state park ownership lie immediately south and northeast of Karner Blue range.

METHODS

Transect butterfly surveys were done annually since 1988, intensively so since 1990, on similar routes over the years that cross (not follow) edges of management treatments and ecotones. A new unit (or subsite) was designated within a site whenever the habitat along the route varied by most recent management (type and timing), habitat degradation, and/or vegetation type, so that the vegetation in each survey unit was relatively uniform (although less so in savanna sites, which have more heterogeneous habitat). Canopy cover of woody plants was estimated informally. Each unit's weather was coded as good, intermediate, or poor for butterfly observation with a graduated scale consistently applied to all units based on objective weather measures. Surveys were timed especially to sample these species of conservation interest: Frosted Elfin (Incisalia irus), Karner Blue, Regal Fritillary, Poweshiek Skipperling (Oarisma poweshiek), Dakota Skipper, Ottoe Skipper (Hesperia ottoe). Vegetative classifications are based on site descriptions for preserves and Curtis (1959). Descriptions of habitat associations are based on the vegetation within about 15 m of butterfly locations. Detailed methods are in Swengel (1993a,b, In review).

RESULTS

During 382 hr and 651 km of census effort in 1,413 units at the 93 prairie sites 1988-93, we counted 80,906 individuals of 90 butterfly species, including 270 Dakota Skippers and 5,509 Regal Fritillaries. Surveys in barrens 1988-93 totaled 159 hr and 302 km and included 5210 Karner Blues. The butterflies varied considerably in habitat association and biogeography, with clinal rather than discrete relationships along these gradients: grassland-forest, wetland-forest, and wetland-grassland (Table 1). Especially but not only in the heterogeneous barrens sites, a species might consistently occupy a particular vegetative type even though that habitat patch seemed relatively small and was surrounded by other habitats.

Butterflies of conservation interest that inhabit herbaceous habitat patches show a pronounced and long-term decline following typical preserve management with frequent burning of the herb layer (Swengel 1993a,b). Un-intensive cutting (mowing, haying, brushing, timber cutting) is more effective at maintaining an abundance and diversity of these species. Limited tests of light prairie grazing show it may also serve rare butterflies better than fire. Low intensity forestry practices, mechanical cutting, recreational uses (e.g. ATV trails), localized burning, and long-term non-management have all been compatible with large populations of Karner Blue (>100 individuals and >60 individuals/hour observed on any single survey 1990-93) (Swengel 1993a). Other specialists in barrens also concentrate in longer unburned areas of fire-managed sites, especially adjacent to areas not managed with fire. They are often more abundant in places with other types of low intensity management maintaining native herbaceous habitat patches. Where fire appears compatible with maintaining habitat occupied by rare prairie-savanna butterflies, it occurred infrequently (interval >15 years) in the context of large adjacent unburned and occupied habitat patches with alternative low intensity management (e.g. timber cutting) (Swengel 1993a, In review).

DISCUSSION

These results agree with butterfly conservation experience around the world, particularly in Europe and Australia (Butterflies Under Threat Team 1986; Kirby 1992; New 1993). Simply preserving habitat is not sufficient to con- serve insect biodiversity. Suitable managements or land uses must also be compatible with the habitat's native biodiversity. Often invertebrates are more sensitive to subtle habitat characteristics and managements than the plants. The plants may be successfully conserved without maintaining the associated animals, but it is impossible to maintain the associated animals successfully without the plants. Increasing the abundance of certain plants may not benefit associated insects, either because the methods to achieve this plant result are inviable for the insects or because the insects are disfavored by the resulting habitat structure.

To "hedge bets" on the often unforeseeable consequences of particular management approaches, diversity and patchiness of management appear desirable both within and among sites, rather than broadly applying a single management strategy for a particular habitat everywhere (Hobbs and Huenneke 1992). Whether a site is managed specifically to conserve insects, declines and extirpations of insects specialized to the habitat indicate that ecological degradation has already occurred there, while maintenance of these species indicates success in ecosystem conservation. These resources (Butterflies Under Threat Team 1986; Kirby 1992; New 1993) provide many useful recommendations for managing ecosystems beneficially for the large invertebrate component of biodiversity. The following recommendations are especially important:

• Diversify types of management as much as possible.
• Never apply a single treatment of any type to most/all of a habitat at one time.
• Burn as little as possible; use alternatives as much as possible.
• Do not burn on a rotation faster than five years.
• Distribute the land area and various habitat types of a preserve as evenly among management units as possible.
• Avoid managing contiguous units in consecutive years.
• Avoid large, abrupt treatments; make them gradual, smaller, patchy.

ACKNOWLEDGMENTS

Scott Swengel was invaluable in the field, in analysis, and in literature review. I thank Lawrence Gall, Paul Opler, Robert M. Pyle, Dale Schweitzer, and John Shuey for many helpful comments. I am very grateful to funders of this research: Lois Almon Small Grants Research Program, The Nature Conservancy Minnesota Chapter, U.S. Fish and Wildlife Service, Wisconsin Department of Natural Resources, and Drs. William and Elsa Boyce.