\ \ =-‘!% OLYMPIC SALAMANDE n k +, ----____ Olympic North j -_----.__ Dunn’s salamanderoften occurs along creeks and streams. In seepsand headwater habitats, the speciesmay be found in the samemicrohabitat with the Olympic salamander (Nussbaumand others 1983). Dunn’s salamanderdoes not occur in the CascadeRange of Washington (fig. 3). This specieslacks larvae; individuals develop directly from egg to adult. Material and Methods We sampled 59 streams(seefrontispiece) in three physiographic provinces: the Southern Washington CascadeRange, the Oregon CascadeRange, and the Oregon Coast Range (Franklin and Dymess 1973). A similar study was conducted in northwestern California and southwesternOregon (Welsh and Lind, this volume). We sampledsix streamsin each forest age-class(young, mature, old growth) in the Oregon and Washington Cascades.In the Oregon Coast Range, we sampled streamsin 3 young, 10 mature, and 10 old-growth stands.We sampled an additional 20 streamsflowing through managedsecond-growthforests to study the long-term effects of logging (Corn and Bury 1989). Proximity of streamsto establishedterresbial study-areas varied. In the Oregon Cascades,all of the streamsflowed through standsused in the terrestrial community studies (see Gilbert and Allwine, this volume c). Few of the streams selectedin the Oregon Coast Rangeor Washington Cascades were in the terrestrial study areas.Ages of thesestandswere basedon visual inspection of dominant trees and available timber maps (USDA Forest Service and U.S. Departmentof the Interior, Bureau of Land Management).We selected permanent, l- to 2-m wide, first- or second-orderstreams. We adaptedthe habitat variables sampled from standard stream-surveytechniquesof Platts and others (1983). Bury and Corn (in press) and Corn and Bury (1989) describethese techniquesin detail. Briefly, sampling consisted of an intensive searchof 10 m of water and bank along a representative stretch of each stream.We usedstandarizedtechniquesfor measuringwater depth, stream width, percentagegradient, pool-to-riffle ratio (100 percent pool = 1.0; 100 percent riffle = 0). percentageslope of the channel on each side of the stream,and water temperature. 355 TAILED FROG DUNN’S SALAMANDER I- We estimatedby eye the particle size of the subsaatum that was most abundant across the stream at ten l-m intervals. We categorized substrataas silt or sand (~2 mm diameter), gravel (2-32 mm), pebble (32-64 mm), cobble (64-256 mm), and boulder (>256 mm). When possible, we recorded where eachanimal was fast seen (pool, riffle or bank-seep)and measuredthe depth of water and size of the cover object over individual animals. All animals were collected by hand. Representativevouchers of each specieswere retained from many sites and deposited at the National Museum of Natural History in Washington, DC, the Burke Museum at the University of Washington, or the Conner Museum at Washington StateUniversity. We identified significant differences in streamcharacteristics among forest age-classesand among provinces with a oneway analysis of variance, which we also used to test for differencesin amphibian densities among age-classes.Densities were transformed as the natural log of (density + 1). and percentagevariables (streamgradient, pool-to-riffle ratio) were arcsine-transformed. Results and Discussion Scopeof Comparisons Considerable variation was found in the biotic communities both within and among streamsin each province we studied, yet we sampleda relatively small number of streams(minimum N = 18 in eachprovince). Thus, our results should be viewed as a first attempt to compare aquatic amphibian communities in the Pacific Northwest. Naturally Regenerated Stands Stream characteristics-The streamswe sampledin young, mature, and old-growth forests had similar gradients, widths, depths, side-slopegradients, and water temperatures(table 1). This finding was somewhatsurprising considering the large geographic area and different mountain ranges where streams were examined (seefrontispiece). We purposely selected headwatersand small streamsthat were about 1 to 2 m wide, however, and some variables (such as water depth) may simply reflect or be correlated to streamsize (width). Sizes of substratawere evenly distributed (P = 0.27) in streamsof young, mature, and old-growth forests (fig. 4). Table l-Mean valuesof physicalcharacteristicsof streams sampledduring streamsurveys Oldgrowth Mature Variable Numberof Streams 22 22 Gadicnt (9%) Width h) 11.4 1.23 14.4 1.16 Pm1 ratio Side slope ($0) water temperature (‘C) 0.38 43.4 to.9 0.40 44.1 10.9 Depth&i) 4.8 4.6 Young P 15 8.3 1.17 4.5 0.40 46.9 11.4 0.23 0.85 0.81 0.83 0.82 0.73 Table 2-Mean density (IndividualsIm~ of amphibiansin streamsfor different forest age-classes Species Tailedfrog Old-growth Mature 22 0.86 22 0.98 1.71 22 0.37 0.73 22 1.39 1.70 1.14 22 0.61 1.09 22 1.21 8 smams Mean 0.13 0.38 16 0.21 0.26 22 2.89 2.74 SD streams Mea” Young P 15 0.84 1.03 15 0.15 0.25 15 1.26 1.48 8 0.37 0.49 9 0.10 0.27 15 2.50 1.92 8 0.50 0.61 16 0.36 0.60 22 3.19 2.50 0.98 0.23 0.93 0.28 0.30 0.62 Siltkand Pebble Gravel Cobble Boulder Substrate Amphibian abundance and forest age-Abundance (individuals/m’) of each amphibian specieswas highly varied, but no significant differences in abundancewere found among any of the forest types (table 2). Giant salamanderswere the most common amphibians. Pacific giant salamanderswere present in 50 of the 59 (85 percent) streamswe sampled,and Cope’s giant salamanderoccurred in 11 of 24 (46 percent) streamswithin their limited geographic range. Together, these specieswere present in 53 of the 59 (90 percent) streamssampled.Tailed frogs were the next most frequent species,occurring in 45 of 59 (76 percent) streams.Olympic salamandersdo not occur in the Mount 0 1 2 3 4 Number of species 357 Table S-Depth of water and size of rocks used for cover for each specks of stream smphiblsn (adult includes all transformed juvenlks and sexually mature adults) cover size (Cm*) Depth (4 A% N Mean SD N Mean SD Adult lava 93 491 4.1 4.9 2.9 4.1 60 302 1.011 415 3,579 668 Adub Olympic salamander Larva 36 170 3.9 3.6 8.3 4.1 z 629 467 637 717 Adult Larva 31 728 5.4 6.8 5.9 6.2 24 347 2.410 712 6,072 2.032 Cope’s giant Larva salamander 76 8.0 7.5 37 807 699 lhnnk salamander Adult - 85 1.003 1,399 Sue&s Taikd frog Pkfi~ giant rakmdcr Species - Rainier area, but we found this speciesin 31 of 55 (56 percent) streamswithin its geographic range. Dunn’s salamander occurred in 23 of 41 (56 percent) possible streamsin the Oregon Cascadesand Coast Range. Two streamslacked amphibians, but most (88 percent) had 2 to 4 spies present (fig. 5). Thus, almost all permanent streamsflowing through the natural forests of the Pacific Northwest have multispecies communities of aquatic amphibians. Microhabitat association-We found no significant relations between the abundanceof any amphibian speciesand the physical habitat variables we measured(streamgradient, pool ratio, mean depth or median subs&atevalues). Significant differences (P < 0.05) were found among speciesin their use of microhabitats within streamswe studied (fig. 6). however. Both tailed frogs and Olympic salamanderswere found most often in riffles, and about 25 percent of the Olympic salamanderswere taken on streambanks or in shallow seepsat streammargins. Cope’s giant salamandermostly frequented pools. but Pacific giant salamanderswere taken about equally in riffles and pools. Dunn’s salamanderis often semi-aquatic, and OUTcapturesof this specieswere almost entirely on stream-bankor seephabitats. The majority of amphibians were found under rock or log cover (we were unable to determine the exact undisturbed position of all animals). Of those capturesthat provided data on the position of animals, 471 of 722 (65 percent) Pacific 358 - - giant salamanders,418 of 564 (74 percent) tailed frogs, 184 of 239 (77 percent) Olympic salamanders,65 of 78 (83 percent) Cope’s giant salamanders,and 94 of 101 (93 percent) Dunn’s salamanderswere collected from under a cover object. Larval giant salamanderswere often found in open water in pools, and tailed frog tadpoles were observedattachedto rocks in riffles by meansof their suctorial mouth. Differences were found between amphibian age-classes(larvae versus all transformed individuals) and among speciesin the depth of water and the size of rock cover under which they were found (table 3). The larvae and neotenesof tbe Pacific giant salamander(P = 0.056) and tadpoles of the tailed frog (P = 0.015) were in deeperwater than adults of each species.No difference was observed between the depth of occurrencefor larval and adult Olympic salamanders(P = 0.44). We found that the larvae of both the Pacific giant salamander(P < 0.001) and tailed frog (P = 0.001) were under smaller rocks than were adults, but no significant differences were found between the use of rocks by larval and adult Olympic salamanders(P = 0.11). Both adults (P = 0.035) and larvae (P = 0.001) of four speties (excluding Dunn’s salamander;table 3) were observed at different water depths. Differences in the size of rocks used for cover among five species(table 3) were also significant for both adults (P = 0.029) and larvae (P < 0.001). Many of thesedifferences in the selection of cover objects likely reflect the marked variation in speciessize. Large adults of the Pacific giant salamanderand adult tailed frogs were mostly found under larger rocks or boulders, whereas the smaller larvae of Olympic salamandersand tailed frogs often occurred under small rocks or in beds of gravel, pebble, and cobble. Table 4-Physical characteristics ot streams in each province Table 5-Mean densities (inditiduals/m2) of stream amphibians in each province; subprovinces are listed from northern to south- ern (seefrontispiece) 18 19.1 1.10 4.2 0.38 48.5 10.9 IS 8.8 1.20 4.6 0.42 38.8 9.7 o.oQ, 0.45 0.42 0.63 0.31 0.002 ongon coast Range 03) Coast (4) Siuslaw (5) 0.76 0.86 0.76 coquille (6) 0.65 Umpqua (8) /_ Andrews (6) $,,Roguem 2.28 1.94 2.41 2.28 2.40 - 0.41 0.18 0.24 0.57 (1.49 : 0.25 0.48 0.88 : 0.38, / 0.10 1.40 1.75 0.04 0.07 0.07 0.28 : : 0 0.15 0.15 0.40 0.40 0.31 0.22 0.83 0 0.29 - 0.81 Qt&m Cascade%:; m : Mt.Hmd(@ i 50 0.29 0.14 0.16 0.22 0.58 ‘_ I 0.01 0.02 so. wasbiigum Substrate Fire 7-Distribution of sizes of substrates in sueatns in the sotiem Washingtan Cascades (SW). the Oregon Cascades(Oc). and the Oregon Coast Range (CR). Regional Patterns Stream characteristics-Analysis of habitat variables suggestedthat tie s!xams we sampled in eachprovince were similar. No significant differences were found in width, depth, or pool ratio among pmvinces (table 4). Streams sampled in the Oregon Cascadeshad steepergradients than in the other regions (P < 0X01), mostly becauseof a few streamswhere very steepreaches(30-45 percent) were sampled.The mean water temperaturein Coast Range streamswas wanner, perhapsreflecting differences in regional climate. Sites in the interior parts of the Coast Range are in areasthat are subject to higher summer temperaturesthan are streamsat higher elevations in the CascadeRange. Cascades (18) ML Rainier (4) Cowlilz River (4) 1.72 0.60 0.42 0.35 1.I 1 - 0.10 Lewis River (5) 0.70 0.45 0.34 0.10 1.20 Wind River (5) 4.32 0.97 0.16 - For OUTsamples,statistically significant variation was found among provinces in substrateof streams.More boulders and fewer pebble or cobble-sized rocks were found in the streams we studied in the Oregon Cascadesthan in the other two regions (fig. 7). Someof the.Oregon Cascadestreamsmay have been underlaid with bedrock, and the reduction in cover sites may partly explain the generally low abundanceof amphibians in someof these streams.Streamswe examined in the Washington Cascadeshad more gravel, pebble, and cobble than those in the Oregon Cascades.The Coast Range sites had more cobble and pebble substratethan either of the Cascadeprovinces, but the differences were minor. Amphibian abundance-Considerable variation was found in the abundanceof each speciesamong the streamswe sampled in the three provinces (table 5). Pacific giant salamanders and Dunn’s salamanderswere most abundant in the streamswe studied in the Oregon Coast Range, and tailed frogs reached their highest densities in streamswe sampled in the Washington Cascades.Olympic salamanderswere more abundant in the streamswe searchedin the Washington Cascadesthan in those we sampledin the Oregon Cascades or Coast Range. Within each of the provinces, someclinal patterns were found. Abundance of amphibians in the Oregon Cascadesites varied fmm the southernmoststreamsthat contained only three speciesin low numbers to streamsin the northernmost area (Mount Hood) that had up to five speciesin moderateto 359 Table 6-Mean bfomass (g/m”) and dens&y (inditidualsh~ of streamamphibiansand salmonid tlshes(from Platts and McHenry 1988)in the Pacillc Northwest Managed Stands Timber harvest increasesinsolation on waterways and raises streamtemperatures(seeBeschta and others 1987, Hartman and others 1987). Such effects usually last only until the canopy is reestablished.In the Oregon Coast Range, Antis and Froehlich (1988) reported that shading over streamsmay reach 50 percent in less than 5 years after clearcutting and may approachprelogging cover by age 10. Tailed fmgs and Olympic salamanders,however, have low and narrow temperature tolerances,and these speciesare likely to be negatively affected by the increasedwater temperaturein smeams in clearcuts (Bury and Corn 1988b, de Vlaming and Bury 1970). Conversely, Pacific giant salamandersmay be more abundant in streamstraversing clearcuts than in densely high abundance.Except for Cope’s giant salamander,amphib- forested stands(Hawkins and others 1983, Mtuphy and ians in Washington streamsshowed a reverse pattern with the Hall 1981, Murphy and others 1981). possibly becauseof most speciesand highest abundancein streamsto the south. enhancedpopulations of invertebrate prey for several years We found few differences from north to south among the after timber harvest. Invertebrates, salmonids and salamander streamsin the Oregon Coast Range, except that Olympic populations decline once shadeis reestablishedover streams salamanderswere more abundant in the southernmoststreams in western Oregon (Hawkins and others 19X2,1983; Murphy (Coquille River drainage). The four coastal streamsin the and others 1981). Coast Range did not differ from inland streams. Logging can causesedimentation in streams,which eliminates Someof thesepatterns may be due to regional variation in crevices and cover habitat apparently neededby larval salaclimate or streamproductivity. Becauseof relatively small manders (Bury 1988b, Bury and Corn 1988b. Hall and others sample sizes,however, observed differences may also reflect 1978). This habitat feature is the most important determinant variations in the collecting efficiency of field crews, or simof the number of amphibians over the long term (Corn and ply random variation among the streamssampled.Further Bury 1989). work is neededto document possible clinal changesin the occurrenceand abundanceof species. Coarsewoody debris in and along streamsreducesdebris torrents and channelization, functions as a sediment nap, Comparison to Salmonid Fishes and provides sourcesof nutrients and cover for animals in Recently, Platts and McHemy (1988) summarizedthe known streams(Franklin and others 1981, Harmon and others 1986, literature on density and biomass of salmonid fshes. For S&l1 and Swanson 1984). Timber harvest can also reduce streamsin the Pacific region, our data indicate that aquatic the introduction of new pieces of down wood into the streamamphibians are 10 times more abundant with 4 times the bed (Bryant 1985, &dell and Swanson 1984, Swansonand biomassthan what was reported for salmonids (table 6). We Lienkaemper 1978), and such material helps maintain the do not have comparabledata for other fishes (for example, ecological role of streams(Meehan and others 1977, sculpins). Even if tailed frogs are excluded becausetheir Scrivener and Andersen 1984). The juxtaposition and patlarvae are herbivorous and the adults forage above or out of tern of stand types resulting from forest fragmentation the water, small streamsstill harbor populations of aquatic becauseof logging, furthermore, may influence the occuramphibians that exceed any known estimate for coldwater rence and abundanceof stream amphibians. Corn and Bury fishes. Several reamns may account for these differences. (1989). for example, suggestthat the presenceof uncut timber upstreaminfluences the presenceand persistenceof Amphibians may reach high densities and biomassbecause aquatic amphibians in streamsflowing through logged areas their adults can exploit both aquatic and terrestrial prey (Bury downstream. and Martin 1973,Nussbaum and others 1983). Salmonid fishes forage on invertebrates that land on or occur in water. The responseof aquatic amphibians to clearcutting or other Fish are attack predators and highly active, but amphibians disturbancesmay vary regionally. Hawkins and others (1988) often hide under cover as sit-and-wait predatorsand thus found that tailed frogs were virtually extinct in drainage convert more energy to biomass than do fishes. Amphibians basins deforestedby the eruption of Mount St. Helens, appearto be the predominant vertebrate predator in many Washington, whereasfrog densities were high in partially or headwatersand small streamsin the Pacific Northwest. completely forested basins. In the Oregon Coast Range, Corn and Buy (1989) reported that the densities and biomassof amphibian species in unlogged streams were 2 to 7 times greater than in logged streams. They suggested that logging severely reduces microhabitats for larval salamanders because streams in logged stands had more fine sediments than control streams. Tailed frogs and Olympic salamanders were especially sensitive to logging, and in some streams these species may experience local extinction immediately after clearcutting (Corn and Bury 1989, Nussbaum and others 1983). Tailed frogs and Olympic salamanders are closely tied to the stream habitats they occupy. These species appear to have limited dispersal abilities that may prevent them from recolonizing altered streams even after the forest canopy has been reestablished. The likelihood of recolonization by tailed frogs is very low in streams surrounded by hot, arid environments (see Daugherty and Sheldon 1982, Hawkins and others 1988, Metter 1967). There are fewer data on the response of aquatic amphibians to timber harvest in the Cascade Range, on the Olympic Peninsula, or in British Columbia. Aquatic amphibians in these regions may have a higher probability of surviving the effects of clearcutting because northern and higher elevation montane regions are generally cooler than those at lower elevations or lower latitudes. The response of Pacific giant salamander populations to logging is dependent on stream gradient (Corn and Bury 1989, Hall and others 1978). For several decades after logging, their population numbers remain small in low gradient streams, whereas they appear to be little influenced in high-gradient streams. Logging does not appear to cause local extinction of this species, and timber harvest may even temporarily enhance some giant salamander populations where there are increases in stream productivity related to greater insolation. I Research Needs and Management Recommendations Additional research is needed on: l l l The effects of timber harvest on the occurrence and abundance of aquatic amphibians in the Pacific Northwest; The habitat preferences and environmental tolerances of aquatic amphibians, especially the Olympic salamander; The life histories, ecology (particularly dispersal abilities), and viable population sizes of tailed frogs and Olympic salamanders; l The effectiveness of different types and sizes of buffer strips along streams to protect amphibians; and l The value of different-sized patches of forested habitat in watersheds as reservoirs or source areas for aquatic amphibians. We recommend that intensive research and improved streammanagement practices be directed at the Olympic salamander and the tailed frog, which we consider to be the species most sensitive to the effects of timber harvest. The survival of aquatic amphibian populations in most forested areas of the Pacific Northwest ultimately will depend upon their abilities to persist in streams flowing through managed forests. Comprehensive planning and the protection of small streams during logging are essential components of management strategies aimed at protecting aquatic amphibian populations. Bury and Corn (1988b) suggested that headwaters should be surveyed for Olympic salamanders and tailed frogs before timber sales, and if these species are present, then protection efforts should be made. Buffer strips have been shown to be effective at protecting stream biota and habitat by maintaining shade and reducing sedimentation (Beschta and others 1987, Hartman and others 1987, Murphy and others 1981, Newbold and others 1980, Raedeke 1988). Costs of buffer strips are higher on smaller streams than on larger streams (Andrus and Froehlich 1988), and it may be difficult to convince land managers that the benefits of buffer strips, and riparian zones outweigh the costs (Bury and Corn 1988b). We therefore need to investigate alternatives for reducing the costs of buffer strips and yet ensure the protection of amphibians, fishes, and other wildlife along headwaters. Retention of deciduous vegetation (bigleaf maple, alder) and unmerchantable conifers (small or cull trees) may be one inexpensive way to preserve shade. We strongly recommend that natural woody debris (large pieces that are partially buried or decayed) be retained in streams. If merchantable timber is felled away from the stream, the deciduous trees within riparian areas would be left mostly intact. Cull or broken trees that fall across streams or along the streambank during timber harvest provide needed sources of nutrients, sediment traps, shade and cover for wildlife. Moderate amounts of slash and coarse woody debris are best left where they fall. Retention of streamside trees and woody debris also reduces the cost of removing logging debris. 361 Aquatic amphibians have small home ranges, and the protection of relatively small patches of streamside habitat might provide habitat for viable populations. In the Oregon Coast Range, we found that some species can occur in disturbed reaches if protected waters are present upstream in the same drainage (Corn and Bury 1989). The protection of only part of a drainage (for example, a forest patch containing headwater creeks) or a buffer zone (for example, a 50-m wide strip) may ensure survival of aquatic amphibian populations. Headwaters and streams comprise less than 5 percent of the total surface area of northwestern forests (Bury 1988). Wise management of this rare and ecologically important resource, however, is essential for maintaining the health and stability of wildlife and fisheries populations in the Pacific Northwest. 362 Acknowledgments We thank Robert M. Storm, and two anonymous reviewers for their comments on the paper. Andrew B. Carey assisted with development of the sampling design, and participated with work in the Oregon Coast Range. We also thank the many biologists who assisted with the field work and stream surveys. This paper is contribution 127 of the Wildlife Habitat Relationships in Western Washington and Oregon Research Project, Pacific Northwest Research Station, USDA Forest Service. q Continue