METHODS

Study Area

North Cascades National Park Service Complex (NOCA) is comprised of North Cascades National Park, Ross Lake National Recreation Area, and Lake Chelan National Recreation Area, and is located in the Cascade Range of northern Washington, USA. There are 156 lakes of interest to fisheries managers in NOCA. All of these lakes are low in chemical ion concentrations and are considered oligotrophic. All but one of these lakes were thought to be historically devoid of fish (Jarvis 1987). Many NOCA lakes were stocked during this century with trout, primarily Oncorhynchus clarki and O. mykiss, to provide recreational angling opportunity.

Fish

Lakes were grouped into three categories: fishless lakes, lakes with non-reproducing trout, and lakes with reproducing trout. In NOCA fish densities in lakes with reproducing trout are generally much higher than in lakes in which trout do not reproduce (Liss et al. 1995). Average fish density, estimated by mark-recapture in nine lakes with reproducing trout, was 524 fish/ha for fish > 177 mm total length (range 250-724 fish/ha except one lake at 98 fish/ha; Liss et al. 1995; Gresswell et al. 1997).Lakes in which fish do not reproduce are periodically stocked with fry at low densities. Average density of trout fry stocked from 1976 to 1993 in 37 high-elevation lakes ( 1100 m) was 179 fish/ha (range 60-375 fish/ha) and the average interval between stocking was > 5 years (Liss et al. 1995). Lack of reproduction in lakes with fish was determined from NOCA stocking records (Jarvis 1987) and field observations (e.g., failure to observe fry or smaller fish, little variation in age and size structure of captured fish, and lack of suitable spawning habitat). Presence or absence of trout in study lakes was verified by gill netting, angling, snorkeling, and observations from shore.

Salamander Density

From 1990 through 1994, larval salamander populations were sampled in 20 fishless lakes (NE), 7 lakes with non-reproducing fish (NRF), and 18 lakes with reproducing fish (RE). We sampled lakes from mid-June to mid-September each year, the period of time in which lakes are typically ice-free.

Because of the remoteness of NOCA lakes, relatively short ice-free periods, periodic wildfires, and periods of inclement weather, sampling frequency of lakes varied within and among years. Seventeen lakes (7 NE, 3 NRF, 7 RE) were sampled 2 or more times in at least 1 year, 8 lakes (2 NE, 1 NRF, 5 RE) were sampled once a year over 2 or more years, and 20 lakes (11 NE, 3 NRF, 6 RE) were sampled only once.

Larval salamanders were censused by snorkel surveys. Because of the remoteness of lakes, snorkeling methods provided the best estimates of larval densities given time and equipment constraints. Snorkel methods tend to underrepresent small, cryptic, and benthic individuals in density estimates (Helfman 1983). Therefore, larval salamander density estimates from snorkel surveys are conservative estimates.

From 1990 to 1993, surveys were conducted during daylight. During these surveys, termed "search surveys," a snorkeler carefully searched through substrate materials (i.e., talus, woody debris, fine organic material, and aquatic vegetation) within 2 m of the shoreline and recorded the number of larvae observed. The length of shoreline surveyed was determined following completion of each search survey.

During 1994, four 25 m segments of shoreline were randomly chosen along the perimeter of each lake. The same segments were surveyed on all sampling visits. Each segment was snorkeled along two transects parallel to shore (e.g., Taylor 1983). One transect was approximately 2 m from shore, and the other transect was over deeper water approximately 5 m from shore. During 2 m surveys, the snorkeler counted larvae within an area extending from the shoreline to approximately 2 m offshore. During 5 m surveys, the snorkeler counted larvae approximately 1.5 m to each side of his or her longitudinal axis. The snorkeler did not disturb substrate material during these surveys. Search surveys also were performed along the same segments of shoreline. To determine if the number of larvae observed was different between day and night, each transect was surveyed during mid-afternoon and approximately 30 min after sunset with the aid of hand-held divelights. When multiple surveys were performed at a lake, salamander density estimates were averaged for each survey technique. Salamander densities were expressed as number of larvae observed per 100 m of shoreline.

Salamander Stomach Contents

Stomachs were collected from 13 salamander larvae from 3 fishless lakes (MR2 [N=9], MR3 [N=2], and MM6 [N=2]) in 1990 and 1991. In the field stomachs and their contents were preserved in 95% ethanol. Organisms found in the stomachs were identified to the lowest possible taxonomic level in the laboratory using a stereomicroscope for benthic macroinvertebrate taxa and an inverted scope at 100X magnification for crustacean zooplankton taxa.

Salamander Behavior

During 1993 and 1994 search surveys, the total lengths (mm) of individual salamander larvae observed during snorkel surveys were visually estimated. Visual size estimates were aided through the use of plastic metric rulers. The number of larvae in each of three size categories (10-30mm, 31-60 mm, and >60 mm) was determined. Larvae also were categorized according to whether they were hidden or were not hidden in substrate materials (e.g., talus, woody debris), in rock crevices, or among dense vegetation. A larva was classified as "not hidden" if a major portion of its body was visible to the snorkeler prior to the search through substrate materials.

Salamander Species Identification

There are only two species of ambystomatid salamanders (A. macrodactylum and A. gracile) in NOCA and they rarely co-occur in a lake (Liss et al. 1995). In NOCA A. gracile is restricted to low elevation lakes on the west slope of the Cascade Range (Liss et al. 1995).

Salamander larvae were captured using hand-nets during snorkel surveys. Captured larvae were taken to a laboratory and reared to metamorphosis to confirm species identification. When it was not possible to transport larvae from the field, larvae were determined to be A. macrodactylum based on larval characteristics (Nussbaum et al. 1983; Leonard et al. 1993); absence of large larvae (>60 mm total length) or neotenes and egg masses characteristic of A. gracile; and the presence of pre-metamorphic individuals with adult coloration.

Lake Physical and Chemical Characteristics

Eleven abiotic variables were measured for each lake. A hand-held sonar gun was used to determine maximum depth of each lake. Lake elevations were derived from 7.5' USGS topographical maps, and lake surface areas were determined by digitization of lake shorelines outlined on these maps. Each time a lake was sampled, water temperature and water chemistry samples were collected at 1 m below the lake surface over the lake's deepest point. Water chemistry samples and temperature recordings were gathered over each lake's deepest point to standardize our sampling; thus enabling us to compare measured variables between lakes. Water samples were collected with a 1.5 L van Dorn sampling bottle. Temperature measurements were determined during mid-afternoon using an Omega 871 thermo-couple. Frozen filtered and unfiltered water samples were transported to the Cooperative Chemistry Analytical Laboratory at Oregon State University, Corvallis, for analyses of total phosphorus, total Kjeldahl-N, ammonium-N, and nitrate/nitrite-N concentrations, and alkalinity, pH, and conductivity.

Zooplankton and Nearshore Benthic Macroinvertebrates

Each time a lake was sampled, crustacean zooplankton were collected using a 20-cm-diameter number 25 (64 µm mesh) zooplankton net. From 1990-1993, three replicate vertical tows were collected in each lake on each sampling occasion. Only one vertical tow was performed on each visit to three lakes sampled in 1989. For each vertical tow, the net was lowered to within 1 m of the lake bottom near the deepest point in each lake ad retrieved upward at a constant rate. In the field samples were preserved in 5% neutral sugar formalin solution (Haney and Hall 1973). In the laboratory samples were split using a Folson plankton splitter. Split portions were allowed to settle for 24 hrs and adult zooplankton in these samples were identified to species and counted using an inverted microscope at 100x magnification (Liss et al. 1995). Zooplankton densities were expressed as number/L.

Benthic macroinvertebrates were sampled using a 17-cm-diameter metal sampling tube (Hoffman et al. 1996). Three samples of each major substrate type in the lake nearshore areas were collected. The tube was placed in position over each sampling site and depressed into the substrate. Material was extracted from the tube to an approximate depth of 5 cm and placed into a 250 µm sieve (U.S.A. Standard Tyler No. 60). Material in the sieve was rinsed with water removed from the tube with a large-bore pipette. The material was placed into a plastic container ad handpicked for organisms. All organisms were preserved in 70% ethanol. In the laboratory organisms were identified to the lowest taxonomic level possible using a stereomicroscope. Macroinvertebrate densities were expressed as number/m².

Statistical Analyses

Statgraphics versions 6.0 and 7.0 were used for all statistical analyses. Each statistical test was performed at = 0.05.

To test for differences in average salamander densities between survey techniques (i.e., search, day 2 m, day 5 m, 2 m, ad night 5 in), a Friedman's F-test was performed on salamander densities in fishless lakes with lakes as blocks and survey techniques as treatments.

Only lakes with at least two zooplankton samples in a given year were used for zooplankton analysis. Mean densities for each taxon for each year were calculated. If lakes were sampled over several years, densities were averaged for all samples. Average densities of benthic macroinvertebrate taxa were calculated in the same manner. Pearson correlation matrices were developed to identify significant relationships between selected abiotic variables, zooplankton densities, and benthic macroinvertebrate densities.

Multiple regression was used to determine relationships between abiotic factors and larval density for each fish category (fishless, non-reproducing fish, and reproducing fish). The dependent variable for regression was the natural logarithm of average larval density in each lake, calculated from search surveys. Values for each chemical variable were averaged over all years in which snorkel surveys were performed. Water temperature averages were calculated from temperature measurements recorded on the day of snorkel surveys. A Pearson correlation matrix was developed to examine relationships among abiotic variables used for multiple linear regression analysis. For the correlation matrix, a sequential Bonferroni adjustment was performed on to eliminate type I error and maintain table-wide significance at = 0.05 (Miller 1981; Rice 1989).

To test for differences in larval densities between fishless lakes and lakes with reproducing ad non-reproducing trout, mean larval density and the 95% confidence interval (CI) were determined for each fishless lake from the linear regression model with TKN concentration and lake elevation as independent variables. Mean larval densities ad 95% CIs for lakes with reproducing fish ad for lakes with non-reproducing fish were also determined. Differences in larval densities between individual fishless lakes and lakes in other fish categories were judged to be significant if 95% CIs did not overlap.

Differences in the proportion of hidden salamanders among larval size classes and among fish categories were investigated using analysis of variance (ANOVA). Proportion of hidden larvae was arcsin-square-root transformed for all ANOVA tests to reduce within-group variability. We tested for differences in proportion of hidden salamanders among larval size classes in fishless lakes. To test for differences in proportion of hidden larvae among fish categories, larvae from all size classes were combined in each fish category and the combined densities were compared between fish categories.