Ecology, 81(5), 2000, pp. 1371±1382 q 2000 by the Ecological Society of America SPATIAL VARIATION IN FISH ASSEMBLAGES ACROSS A BEAVER-INFLUENCED SUCCESSIONAL LANDSCAPE ISAAC J. SCHLOSSER1,3 AND LARRY W. K ALLEMEYN2 1Department of Biology, Box 9019, University Station, Grand Forks, North Dakota 58202 USA 2United States Geological Survey, Columbia Environmental Research Center, International Falls Biological Station, 3131 Highway 33, International Falls, Minnesota 56649 USA Abstract. Beavers are increasingly viewed as ``ecological engineers,'' having broad effects on physical, chemical, and biological attributes of north-temperate landscapes. We examine the in¯uence of both local successional processes associated with beaver activity and regional geomorphic boundaries on spatial variation in ®sh assemblages along the Kabetogama Peninsula in Voyageurs National Park, northern Minnesota, USA. Fish abun- dance and species richness exhibited considerable variation among drainages along the peninsula. Geological barriers to ®sh dispersal at outlets of some drainages has reduced ®sh abundance and species richness. Fish abundance and species richness also varied within drainages among local environments associated with beaver pond succession. Fish abun- dance was higher in upland ponds than in lowland ponds, collapsed ponds, or streams, whereas species richness was highest in collapsed ponds and streams. Cluster analyses based on ®sh abundance at sites classi®ed according to successional environment indicated that four species (northern redbelly dace, Phoxinus eos; brook stickleback, Culaea incon- stans; ®nescale dace, P. neogaeus; and fathead minnow, Pimephales promelas), were pre- dominant in all successional environments. Several less abundant species were added in collapsed ponds and streams, with smaller size classes of large lake species (e.g., black crappie, Pomoxis nigromaculatus; smallmouth bass, Micropertus dolomieui; yellow perch, Perca ¯avescens; and burbot, Lota lota) being a component of these less abundant species. The addition of smaller size classes of large lake species indicates that dispersal of early life-history stages from Kabetogama Lake played a role in determining the species richness and composition of less abundant species in successional environments on the peninsula. Furthermore, collapsed-pond and stream environments closer to Kabetogama Lake had higher species richness than similar successional sites located farther from the lake. Cluster analyses based on ®sh abundance at sites classi®ed according to drainage indicated that species composition among drainages was in¯uenced both by the presence or absence of geological barriers to ®sh dispersal and the nonrandom distribution of collapsed ponds and streams. Based on these results, we present a hierarchical conceptual model suggesting how geomorphic boundaries and beaver pond succession interact to in¯uence ®sh assemblage attributes. The presence of a productive and diverse ®sh assemblage in headwater streams of north-temperate areas requires the entire spatial and temporal mosaic of successional habitats associated with beaver activity, including those due to the creation and abandonment of beaver ponds. The ultimate impact of the local successional mosaic on ®shes, however, will be strongly in¯uenced by the regional geomorphic context in which the mosaic occurs. Key words: beavers; boundary effects; Castor canadensis; dispersal; ®sh assemblages; landscape; Minnesota; pond succession; spatial hierarchy; spatial scale; successional mosaic; Voyageurs National Park, Minnesota (USA). INTRODUCTION chically structured and scale-dependent levels of var- iation and complexity (Allen and Starr 1982, Naveh A fundamental shift has occurred in ecology, incor- porating a broader geographic perspective (Brown and Lieberman 1984). As different levels in the envi- 1987, Wiens 1989, Gilpin and Hanski 1991, Hanski ronmental hierarchy are transcended, variation in dis- and Gilpin 1997). Ecologists are increasingly exam- persal and invasion can affect population sizes, de- ining spatial and temporal variation to understand how mography, and coadaptation, as well as food webs and different processes become important in determining assemblage structures (Tonn et al. 1990, Endler 1991). population dynamics, species interactions, and assem- Furthermore, these hierarchies frequently occur within blage structure as ecosystems are viewed over hierar- landscape mosaics that are normally characterized by dynamic successional patch structure (Pickett and White 1985, Chesson and Huntly 1997, Wiens 1997). Patch quality, connectivity, context, and boundary 3 E-mail: [email protected] characteristics are, in turn, thought to have profound 1371 1372 ISAAC J. SCHLOSSER AND LARRY W. KALLEMEYN Ecology, Vol. 81, No. 5 in¯uences on dispersal and subsequent food web in- Rahel 1984, Magnuson et al. 1989, Tonn 1990, Tonn teractions (Pickett and White 1985, Wiens 1997). Con- et al. 1990, Magnuson et al. 1998). sequently, a fundamental objective for ecology must Our objective was to assess how the shifting suc- be to understand how hierarchically dependent attri- cessional mosaic of local aquatic habitat patches cre- butes of environmental variation interact to in¯uence ated by beaver in¯uences the number, identity, and rel- population dynamics and assemblage attributes in the ative abundance of ®sh species and place this in the context of landscape succession (Hanski and Simber- context of processes operating at the spatial scale of loff 1997). drainage basins. Speci®cally, we ®rst examine variation Modi®cation of aquatic ecosystems by beaver has in ®sh abundance and species richness among drainages resulted in a shifting successional mosaic of aquatic along the Kabetogama Peninsula in Voyageurs National patches embedded within the larger regional landscape Park in northern Minnesota, USA. We then explore how (Johnston and Naiman 1987, 1990, Naiman et al. 1988, this variation relates to successional stages associated Pastor et al. 1993, Johnston 1995, Snodgrass 1997, with beaver activity. Lastly, we examine how species Schlosser 1998, Schlosser et al. 1998, Snodgrass and composition of ®sh assemblages varies among succes- Meffe 1998). Since beaver ponds are created and main- sional environments and drainages. tained by living organisms, they are temporally dy- namic, as stream reaches are colonized, ¯ooded, and STUDY AREA eventually abandoned by beaver. This results in both a The study was conducted along the 294-km2 Kabe- spatial and temporal mosaic of aquatic patches on the togama Peninsula in Voyageurs National Park. Glacial landscape, including ponds, collapsed ponds, and activity 10 000±12 000 yr BP created three large lakes streams (Naiman et al. 1988). Additional complexity surrounding the peninsula; Rainy Lake to the north, occurs when these aquatic environments are placed in Namakan Lake to the east, and Kabetogama Lake to the context of the larger drainage network. Variation the south. Numerous drainages located along the south- in valley shape, for instance, has a strong in¯uence on ern edge of the Kabetogama Peninsula enter into Ka- pond morphology, creating either deep, narrow ``up- betogama Lake, ranging from Locator and Sucker land'' ponds in constrained valleys or broad, shallow Creek in the far northwest to Shoepack and East Shoe- ``lowland'' ponds in unconstrained reaches (Johnston pack in the southeast (Schlosser et al. 1998: Fig. 1). and Naiman 1987). Spatial context within the longi- Two drainages located in the middle of the peninsula tudinal drainage network can also in¯uence the mag- have potential barriers to ®sh movement between them nitude of hydrologic ¯uxes through the ponds, and their and Kabetogama Lake. Clyde Creek exhibits a 5±10 m spatial proximity to more stable aquatic ecosystems drop in elevation near Kabetogama Lake, associated downstream, both of which are likely to impact dis- with a series of small (0.5±1 m) boulder and bedrock persal and exchange processes. (Schlosser 1987, waterfalls. A 1.5±2 m bedrock waterfall is also located 1995a, b, Osborne and Wiley 1992). Finally, consid- at the outlet of the West Lost Ponds drainage. Since erable temporal variation occurs in most north-tem- the early part of this century, increased beaver activity perate aquatic environments, associated with seasonal on the peninsula has transformed the .300 km of ex- and annual variation in temperature and oxygen, with isting stream channel within the drainages (Naiman et the potential for oxygen depletion being particularly al. 1988). Modi®cation of the aquatic ecosystem by severe in small, organically rich ponds and lakes in beaver has resulted in a spatial and temporal mosaic north-temperate areas (Tonn and Magnuson 1982, Ra- of habitats associated with age, stage of succession, hel 1984, Johnston and Naiman 1987, Magnuson et al. and local environment of beaver ponds (Johnston and 1989, Schlosser et al. 1998). Naiman 1987, Naiman et al. 1988). Previous studies of ®sh assemblages in small lakes suggest a combination of factors interact to control ®sh METHODS AND MATERIALS assemblage attributes. In particular, the number, iden- Fish sampling tity, and relative abundance of ®sh species in small lakes can be viewed as the product of a series of ``®l- Fish were sampled at 23 sites in 12 drainages along ters'' ranging from continental to local spatial scales the southern edge of the Kabetogama Peninsula (see (Tonn 1990, Tonn et al. 1990). At regional scales, crit- Appendix