© 2002 Blackwell Science Ltd. http://www.blackwell-science.com/geb RESEARCH LETTER Global Ecology & Biogeography (2002) 11 , 169 –178 Blackwell Science Ltd The endemic headwater stream amphibians of the American Northwest: associations with environmental gradients in a large forested preserve MICHAEL J. ADAMS and R. BRUCE BURY USGS Forest and Rangeland Ecosystem Science Center, 3200 SW Jefferson Way, Corvallis, OR 97331, U.S.A., [email protected], [email protected] ABSTRACT We used a large forested preserve (Olympic National Park, USA)to examine the habitat associations of a unique and environ-mentally sensitive stream amphibian fauna: Ascaphus truei Stegneger, Rhyacotriton olympicus (Gaige) and Dicamptodoncopei Nussbaum. We quantified the relative abundance ofstream amphibians and compared them to physical, topo-graphic, climatic and landscape variables. All three specieswere associated with the south-west to north-east climategradient, tending to be most abundant in the south-west.Although a habitat generalist relative to the other two species, Dicamptodon copei was absent from the north-eastern portionof the park. Ascaphus truei and Rhyacotriton olympicus wereboth associated with coarse substrates and steep gradients.Unlike studies in harvested forests, all stream amphibianswere common in waters with unconsolidated surface geology(e.g. marine sediments that erode easily). Studies of ecologicalpreserves can provide an important baseline for evaluatingspecies associations with environmental gradients and canreveal patterns not evident in more disturbed landscapes. Key words Ascaphus truei , Dicamptodon copei , environ-mental gradients, geology, headwater streams, North America,Olympic National Park, preserves, Rhyacotriton olympicus ,substrate. INTRODUCTION The forests of the Pacific Northwest host a unique and environ-mentally sensitive stream-amphibian fauna (Nussbaum et al. ,1983; Bury, 1994). This endemic fauna includes Ascaphustruei Stejneger (tailed frog), which is the sole member of thefamily Ascaphidae and considered the most primitive extantanuran in the world (Ford & Cannatella, 1993), and the sala-mander families Dicamptodontidae and Rhyacotritonidae.Several workers have described habitat associations for thesespecies but have based their work predominately (Dupuis et al. ,2000) or entirely (Diller & Wallace, 1996, 1999; Wilkins &Peterson, 2000) in harvested forests. Few studies have examinedpatterns in unharvested forests and these have focused on therole of stand age (Welsh, 1990; Bury et al. , 1991). Here, wedescribe habitat associations for stream amphibians in therelatively pristine forests of Olympic National Park (ONP).ONP provides an opportunity for the study of ecologicalpatterns along pronounced environmental gradients. It issituated on the Olympic Peninsula in Washington State, USA,and is surrounded by the Pacific Ocean to the west, theStraight of Juan de Fuca to the north and Hood Canal andPuget Sound to the east (Fig. 1). ONP is 370 000 ha and con-tains the bulk of the Olympic Mountains, which are isolatedby water and low elevation areas from other ranges such asthe Cascade Mountains ( c . 80 km east) and the Willipa Hills( c . 60 km south). The Olympic Mountains rise to 2400 mand produce a pronounced rain shadow: average annualprecipitation ranges from 6000 mm on the south-west side to450 mm on the north-east side. The relative isolation and markedenvironmental gradients of ONP are ideal for describingenvironmental limits and associations. Moreover, almost nologging has ever occurred inside the park’s boundaries, offer-ing a marked contrast with the intensively harvested forests ofmuch of the Pacific Northwest.Three species of stream-breeding amphibian occur in ONP: Ascaphus truei , Rhyacotriton olympicus (Gaige) (Olympic tor-rent salamander) and Dicamptodon copei Nussbaum (Cope’sgiant salamander). All require permanent water for their Corresponding author: Michael J. Adams, USGS Forest and Range-land Ecosystem Science Center, 3200 SW Jefferson Way, Corvallis,OR 97331, USA. E-mail: [email protected] GEB_272.fm Page 169 Saturday, February 23, 2002 3:59 PM170 M. J. Adams and R. B. Bury © 2002 Blackwell Science Ltd, Global Ecology & Biogeography , 11 , 169 –178 multi-year larval stages (Nussbaum & Tait, 1977; Bury &Adams, 1999). D. copei is almost obligately paedomorphic,and so may be especially reliant on permanent waters forsurvival (Nussbaum et al. , 1983). Ascaphidae and Rhyacotri-tonidae are two of the coldest adapted amphibians in theworld (Brattstrom, 1963; de Vlaming & Bury, 1970). A. truei , R. variegatus Stebbins and Lowe and D. tenebrosus Good haveeach been shown to have positive associations with coarsecobble and boulder substrates (Corn & Bury, 1989; Diller& Wallace, 1996, 1999 ) . All decreased in abundance withthe input of fine sediments (Hawkins et al. , 1988; Welsh &Ollivier, 1998). The density of D. tenebrosus increased withthe experimental input of large stones (Parker, 1991). Becauseof their association with coarse substrates and cold watertemperatures, these species may be sensitive to timber harvest(Corn & Bury, 1989; Welsh, 1990; Dupuis & Steventon, 1999),although they persist in many managed forests (Diller & Wallace,1996, 1999; Wilkins & Peterson, 2000).Here, we report results of a 3-year survey that determined thedistribution and abundance of stream-breeding amphibiansacross environmental gradients in ONP. We examined in-streamhabitat features, topographical features and climate gradientsthat might be affected by global change or timber harvest. Wecontrast our findings with those of studies in other parts of thisregion where intensive timber harvest occurs. METHODS We surveyed 141 headwater streams in 12 of 13 major drain-ages in ONP (Fig. 1). Each stream was sampled once duringthe summer months (June–August), 1996–98. Because off-trail access to ONP is difficult, we used trails and a few roadsas a network of nonrandom transects covering the park. Thepopulation of streams from which study sites were chosen wasdetermined by hiking the roads and trails in late summer orfall and recording the location of streams that still had waterflowing. Many of these streams did not appear on maps. Wethen selected randomly one-third to one-half of the streamswithin each drainage that were crossed by main trails androads. The proportion varied due to changes in logistical con-straints.