METHODS

Experiments were conducted in a 10x12 m building at Oak Creek Laboratory of Biology. Air temperature in the building was not controlled. Twelve Rubbermaid brand livestock watering tanks were used as artificial ponds. Each oval tank was constructed of fiberglass and plastic composites and had a capacity of 1136 L (166x152x60 cm). Each tank was filled with well-water to a depth of 38 cm (water volume of each tank was = 616 L). Substrates that simulated those typical of lakes of the northern Cascade range (i.e., talus or rock slide debris, woody material such as submerged tree limbs and branches, submerged vegetation, and open areas) were established in each artificial pond.

Before the addition of water, each artificial pond was filled to a depth of 4 cm with washed sand. Additional substrate materials were placed over the sand bottom in each pond to simulate talus, woody material, and aquatic vegetation. The amount of each substrate type in each pond was roughly equal. To simulate talus, 12 quarry rocks, approximately 2 L each (determined by water displacement), were arranged in each pond in the shape of a rectangular based pyramid (40x60x20 cm). To simulate woody material, 13 tree limbs (seven with 8-10 cm diameters and six with 2 cm diameters) were bound together with nylon cord in the same general shape as the talus and anchored to the bottom of each pond with a concrete pipe (44 cm long and 15 cm diameter). Eight plastic aquarium plants (36 cm high) were attached to plastic disks 30 cm in diameter, which were anchored with sand. Total area for each tank bottom was 16,217 cm² ; wood and rock each occupied 2400 cm², artifical aquatic vegetation occupied 706.9 cm², and open sand occupied the remaining area (10,710.1 cm²).

Collection of Larvae

Recently hatched A. macrodactylum larvae (15-20 mm total length) used in laboratory experiments were collected by dip-net on 16 and 18 April 1996, from a low-elevation, ephemeral pond in Benton County, Oregon. Embryos of A. gracile were collected on 28 May 1996, from Pika Lake in the Cascade mountain range, Linn County, Oregon. Additional A. gracile embryos. were collected on 14 June 1996, from Upper Panther Lake, a lake in the northern Cascade range of Washington, USA.

Prior to the start of each experiment, embryos and larvae were held in 40-L aquaria filled with well-water maintained at 8-11 °C. Larvae were exposed to 12 hr of artificial light each day and were fed live Tubifex sp. ad libitum twice a week.

Experiment Description

Experiments to determine effects of trout on survival and growth of larval A. macrodactylum and A. gracile were performed separately for each salamander species. One trout was placed in each of six treatment ponds while six ponds without trout served as controls. Experiments were performed for 30 d for each salamander species. Rainbow trout (O. mykiss) obtained from Oregon State University, Department of Food Science and Technology, were used for each experiment. Treatment and control ponds were randomly selected.

Twenty larvae were stocked in each artificial pond. Prior to stocking, the snout-vent length (SVL) of each larva was measured. Larvae were randomly assigned to 12 groups of 20 individuals each. Each group of 20 larvae were then weighed collectively and randomly assigned to a treatment or control pond. Forklength (FL; distance from tip of snout to the fork of the tail) and body mass of each trout was recorded [A. macrodactylum experiment, FL mean = 211.7 mm (206-217mm), mean mass = 112.2 g (107-121 g); A. gracile experiment, FL mean = 215.3 mm (212-220 mm), mean mass = 129.2 g (114-146 g)]. One trout was released into each treatment pond 5 hr after larvae had been added.

Artificial light was provided to all tanks from 0700 - 1900 hr each day. All tanks were inoculated with 24 g of Tubifex at least one day prior to the start of each experiment. During each 30 d experiment, 12 g of Tubifex were dispersed throughout each tank three times per week. Tanks were inspected for abnormalities (i.e., water loss, death of larvae or trout, etc.) each day during the experiments.

After each 30 d experiment, trout were removed from treatment ponds and substrate materials were enclosed with mesh screens and carefully dismantled. Number of larvae in each substrate was recorded as substrates were dismantled. Group mass and individual SVLs of surviving larvae were measured. Observations of substrate use by larvae prior to termination of the experiments were not possible since this would have required dismantling substrates while the experiments were in progress. Average daily ambient water temperature at noon was 10.8 °C (9.5-11.6 °C) for the A. macrodactylum experiment and 18.3 °C (16.3-19.3 °C) for the A. gracile experiment. The difference in daily water temperature between experiments was a result of seasonal change in air temperature in the laboratory buildings.

Statistical Analyses

Differences in larval survival between treatment and control ponds were tested by Analysis of Variance (ANO VA) using arcsin square-root transformed proportion of larvae surviving in each pond. Average SVL and average mass of surviving larvae were compared between treatment and control ponds using unpaired Wilcoxin tests. Statgraphics version 7.0 was used for ANOVA's and Wilcoxin tests. Differences in the substrate locations of larvae between treatment and control ponds were tested by generating full and reduced generalized linear models for both species (Ramsey and Schafer 1997). The log-transformed location of larvae in each pond was the dependent variable and the proportions of larvae in each substrate were independent variables for the models. Full models included the binary response variable of trout presence and absence. The binary response variable was excluded from reduced models. Drop in Deviance tests were then used to determine whether fish presence and absence was a significant variable in describing variation in substrate location of larvae (Ramsey and Schafer 1997). Substrate location of larvae was analyzed using SAS version 6.10. All tests were performed at = 0.05. One control pond in the A. gracile experiment was omitted from statistical analyses due to a disease outbreak.