RESULTS

Abiotic Relationships

In fishless lakes in 1994, there were no significant differences in A. macrodactylum larval densities between survey types (Table 5.1; Friedman's F-test, P = 0.34). Differences in larval densities between survey types in lakes with reproducing fish were not tested statistically since virtually no larvae were observed either during the day or at night. In addition, no larvae were observed in four of eight fishless lakes. Although there were no statistically significant differences between survey types, search surveys appeared to provide the most conservative estimates of larval densities.

Table 5.1. Ambystoma macrodactylum larval density estimates (larvae/100 m of shoreline) for fishless lakes (NF) and lakes with reproducing fish (RF) in 1994.

^aLetter and number system for lake identification used by NOCA.
^bNight surveys were not performed (NP) in some lakes.

Stepwise linear regression for all fishless lakes identified statistically significant relationships between In larval density, total Kjeldahl-N (TKN), and elevation (R² = 0.68, P = 0.0001, Figure 5.1):

The relationship with larval density was positive for TKN and negative for elevation. RD3, a low elevation (802m) fishless lake with high larval density, was withheld from a second regression analysis to determine the influence of this lake's elevation in the model. Stepwise regression when RD3 was omitted identified a statistically significant positive relationship between larval density and both TKN and conductivity (R² = 0.6938; P = 0.0001). Thus, when RD3 was omitted, lake elevation no longer had a detectable influence on larval density.

Total Kjeldahl-N had significant positive correlations with total phosphorus, water temperature, and ammonium-N (Pearson correlation coefficients; r = 0.74, P 0.0001; r = 0.51, P = 0.0004; r = 0.46, P = 0.00 17, respectively). Thus, NOCA lakes with higher TKN concentrations tended to have relatively higher concentrations of both total phosphorus and ammonium-N and higher water temperatures.

Salamander Stomach Contents

Benthic macroinvertebrates were identified in 75% of larval salamander stomachs (Table 5.2). The benthic taxon, Diptera, composed primarily of chironomid larvae, was found in the highest proportion of salamander stomachs. Beetle larvae (Coleoptera) and caddisfly larvae (Trichoptera) also formed significant proportions of larval diets. Crustacean zooplankton also were a important component of salamander diets. Cladocerans were the most common zooplankton taxon found in stomachs.

Table 5.2. Percentage of stomachs from 13 Ambystoma macrodactylum larvae that contained benthic macroinvertebrates and crustacean zooplankton taxa.

Crustacean Zooplankton and Benthic Macroinvertebrates

Total crustacean zooplankton density ad cladoceran density were positively correlated with TKN (Pearson correlation coefficients; r = 0.57, P = 0.0053, Figure 5.2a; r = 0.69, P = 0.0003, Figure 5.2b; respectively). Furthermore, there was a significant positive relationship between percent of cladocera composing the crustacean zooplankton communities ad TKN (r = 0.59, P = 0.0040, Figure 5.2c). At low TKN concentrations (0.0 - =0.05 mg/L), zooplankton density was very low and the zooplankton communities were composed almost exclusively of copepods. There was no statistically significant relationship between TKN and total benthic macroinvertebrate density (Pearson correlation coefficients, Figure .3, r = -0.32 16, P> 0.05) or chironomid density (Figure 5.3, r = -0.0899, P> 0.05).

Fish Effects

No significant linear relationships between larval salamander density and any abiotic factor were identified by stepwise linear regression for lakes with reproducing fish or for lakes with non-reproducing fish. When larval densities in lakes with reproducing fish were fitted to a regression model with TKN and elevation as independent variables, neither TKN slope nor elevation slope were significantly different from zero (Figure 5.1; P = 0.68, P = 0.39, respectively). A similar regression analysis of larval densities in lakes with non-reproducing fish also found that TKN slope (P = 0.35) and elevation slope (P = 0.56) were not significantly different from zero (Figure 5.1).

Larval densities in lakes with reproducing trout were low, making detection of significant relationships with abiotic variables difficult. Larval densities in lakes with non-reproducing fish also were low and detection of significant relationships between larval density and abiotic variables for these lakes was further limited by a small sample size (n = 7).

Because multiple regression revealed no significant relationships between larval density and abiotic factors for either lakes with reproducing fish or for lakes with non-reproducing fish, larval densities from all lakes in each group were averaged and the 95% CI for each group average was determined. To assess fish effects on larval density, the 95% CIs for lakes with reproducing fish and for lakes with non-reproducing fish were compared to 95% CIs for individual fishless lakes determined from the multiple regression model with TKN and elevation as independent variables.

For fishless lakes with TKN < 0.045 mg/L, mean larval densities generated by the linear regression equation were not significantly different from mean larval densities in either lakes with reproducing fish or lakes with non-reproducing fish (Table 5.3). However, for all ten fishless lakes with TKN 0.045 mg/L, predicted mean larval densities were significantly greater than in lakes with reproducing fish (P < 0.05). Of the reproducing fish lakes with TKN 0.045 mg/L, none had larval densities that exceeded 1.22 larvae/100m of shoreline. Only four of the ten fishless lakes with TKN 0.045 mg/L had significantly higher mean larval densities than lakes with non-reproducing fish (P < 0.05). These fishless lakes all had relatively high concentrations of TKN (0.095 mg/L) and relatively high larval densities.

Table 5.3. Comparison of 95% confidence intervals for larval A. macrodactylum densities between individual fishless lakes (NF), lakes with non-reproducing fish (NRF), and lakes with reproducing fish (RF).

^aLetter and number system for lake identification used by NOCA.
^bMean larval densities and 95% confidence intervals were determined from a multiple linear regression model using TKN and lake elevation as independent variables.
^cA) indicates a significant difference in larval density between a fishless lake and lakes with reproducing fish, and B) indicates a significant difference in larval density between a fishless lake and both lakes with reproducing fish and lakes with non-reproducing fish.

Behavior

In fishless lakes there were no significant differences among the three larval size classes in proportion of larvae hidden in substrate material (Table 5.4; ANOVA, P = 0.16). Lakes with non-reproducing fish and lakes with reproducing fish were not tested for differences in proportion of hidden larvae between size classes because few lakes in these categories contained all three larval size classes. There were no significant differences in the proportion of hidden larvae among fish categories (Table 5.4; ANOVA, P = 0.50). Although statistical analysis did not identify any significant differences in proportion of hidden larvae among larval size classes or among fish categories, there was a tendency for the percent of hidden larvae to increase as larval size increased in each fish category. Furthermore, there was a tendency for a greater percentage of larger larvae (>30 mm) to be hidden in lakes with both non-reproducing and reproducing fish than in fishless lakes.

Table 5.4. Total number of larval Ambyostoma macrodactylum observed and the number and percent hidden in substrate materials during search surveys of fishless lakes (NF; n=10), lakes with non-reproducing fish (NRF; n=5), and lakes with reproducing fish (RF; n=5) in 1993 and 1994.