Statistical Analysis
We first present a descriptive statistical summary of the spawning fish
count data for lake whitefish and cisco from 2016 to 2019, and for lake
trout from 2016 to 2018, to provide a context for interpreting egg
densities and survival rates pre- and post-habitat enhancement. For the
habitat data, we calculated the mean proportion of substrate types, mean
mid-water column velocity, and mean depth using the data collected in
2016. We compared the habitat characteristics between survey plots with
eggs and those without eggs, as well as between survey plots in Reach 1
and the Tailrace Area. To determine distinguishable habitat features for
plots with eggs, we conducted Analysis of Variance (ANOVA) and tested
for statistical differences between the independent habitat factors and
their interaction effect. Prior to the tests, the habitat data were
transformed (square root of proportions and logarithm of depths and
velocities) to meet the assumption that residuals are normally
distributed. Using the egg data collected from 2017 to 2019, we
calculated the means and standard errors (SEs) for egg densities per
species. We tested the effect of habitat type (Reach 1 Natural, Reach 1
Enhanced, and Tailrace Area), year (2017, 2018, and 2019), and the
interaction effect of habitat type and year on egg densities using a
two-factor permutational ANOVA (Anderson, 2001). Permutational analysis
of variance often provides more power in situations where independence
of variance may be an issue and assessing multiple factors (Anderson,
2001). In cases where the interaction term was significant, post hoc
tests were conducted for each study year and habitat type, with p-values
adjusted using the Holm method to correct for multiple comparisons
(Holm, 1979). Chi-squared tests were also performed to analyze the
differences in the proportion of live eggs between habitat types and
years, with p-values adjusted using the Holm method (Holm, 1979).
Statistical significance was determined using an alpha level of 0.05.
All statistical summaries and analyses were conducted in R.
RESULTS
During the period of September through October, the mean temperature at
Bluefish Lake was 8.6°C, slightly below the long-term average of 8.8°C,
which was calculated based on data recorded over a 30-year period from
1991 to 2020. Among the four study years, 2018 had the lowest mean fall
temperature, measuring 6.1°C. On the other hand, 2017 had the highest
mean fall temperature of 10.4°C. Both 2016 and 2019 had an average fall
daily temperature of 9.0°C. The mean daily discharge was recorded as
21.5 m3/s. It is worth noting that these flows were
below the long-term average of 30.8 m3/s, which was
calculated based on the daily fall flows summarized over a 30-year
period from 1991 to 2020. Among the four study years, 2016 had the
lowest mean daily discharge, measuring 15.1 m3/s.
Subsequently, there was a gradual increase in mean daily discharge, with
values of 18.1 m3/s in 2017, 23.4
m3/s in 2018, and 29.2 m3/s in 2019.