Results

In 1996, for the upper site (3170 m), analysis of the time of flowering showed that the effect of the snowmelt date (date) is significant (p < 0.001), snow removal treatment (treatment) alone is not significant (p > 0.5), and the interaction date x treatment is significant (p < 0.01) (Table 5.1, Fig. 5.1).

In 1997, similar results were found for the middle (2920 m) and the upper sites (3170 m) (Tables 5.2 and 5.3, Figs. 5.2 and 5.3).

Table 5.1. Repeated Measures Analysis. Effect of Snow Removal on Flowering Date (1996, 3170 m).

Source

Effect

LSdf

Ddf

F

P

Treatment

Fixed

1

8

0.16

0.6973

Block

Random

4

Block x Treatment

Random

4

Date

Fixed

12

92

28.50

0.0001

Date x Treatment

Fixed

12

92

4.46

0.0001

Error

Random

92

Two treatments, five blocks, and 13 repeated measures. Treatment and date were treated as fixed effects and block as a random effect. Repeated = date; LSdf = least squares degrees of freedom, Ddf = denominator degree of freedom. Significant effects are shown in boldface type.

160 165 170 175 ISO 185 190 Julian Date iiiiii

160 165 170 175 ISO 185 190 Julian Date

Figure 5.1. Snow removal experiment 1996, 3170 m. Plot of the average frequency of flowers and date (total number of flowers were transformed to a relative frequency from the maximum number of flowers per plot). Julian date 165 corresponds to June 13. Values (± standard error) are based on the mean of five plots.

In 1997 individual plants were followed during the flowering season in the upper site and number of flowers per individual plant was counted every other day. I analyzed the number of days to reach maximum number of flowers, following the first observation day. One-way ANOVA showed that the snow-removal treatment had a significant effect on phenology (p < 0.05) when numbers of days from the first census to the peak of flowering is analyzed. Treatment plants reached their maximum number of flowers on average 11/2 days earlier than the controls.

In summary, 1996 and 1997 experimental data indicate that the time of flowering was earlier on snow removal plots. In addition, flowering duration was longer in the snow removal plots (see middle site, 1997; Fig. 5.2). The snow removal effect on flowering varied significantly across time; difference in the proportion of open flowers between controls and snow removal treatment was significant only at the beginning of the season. In 1997, individual-plant

Table 5.2. Repeated Measures Analysis. Effect of Snow Removal on Flowering Date (1997, 2920 m).

Source

Effect

LSdf

Ddf

F

P

Treatment

Fixed

1

8

5.86

0.0419

Block

Random

4

Block x Treatment

Random

4

Date

Fixed

12

96

49.85

0.0001

Date x Treatment

Fixed

12

96

11.12

0.0001

Error

Random

96

Two treatments, 5 blocks and 13 repeated measures. Treatment and date were treated as a fixed effects and block as a random effect. Repeated = date; LSdf = least squares degrees freedom, Ddf = denominator degree of freedom. Significant effects are shown in boldface type.

Table 5.3. Repeated Measures Analysis. Effect of Snow Removal on Flowering Date (1997, 3170 m).

Source Effect LSdf Ddf F P

Table 5.3. Repeated Measures Analysis. Effect of Snow Removal on Flowering Date (1997, 3170 m).

Source Effect LSdf Ddf F P

Treatment

Fixed

1

8

0.17

0.6892

Block

Random

4

Block x Treatment

Random

4

Date

Fixed

9

72

98.73

0.0001

Date x Treatment

Fixed

9

72

4.40

0.0001

Error

Random

72

Two treatments, five blocks and 10 repeated measures. Treatment and date were treated as a fixed effects and block as a random effect. Repeated = date; LSdf = least squares degrees freedom, Ddf = denominator degree of freedom. Significant effects are shown in boldface type.

Figure 5.2. Snow removal experiment 1997, 2920 m. Plot of the average frequency of flowers versus Julian date (total number of flowers were transformed to a relative frequency from the maximum number of flowers per plot). Julian date 162 corresponds to June 11. Values (± standard error) are based on the mean of five plots.

Figure 5.2. Snow removal experiment 1997, 2920 m. Plot of the average frequency of flowers versus Julian date (total number of flowers were transformed to a relative frequency from the maximum number of flowers per plot). Julian date 162 corresponds to June 11. Values (± standard error) are based on the mean of five plots.

Figure 5.3. Snow removal experiment 1997, 3170 m. Plot of the average frequency of flowers versus Julian date (total number of flowers were transformed to a relative frequency from the maximum number of flowers per plot). Julian date 173 corresponds to June 22. Values (± standard error) are based on the mean of five plots.

Figure 5.3. Snow removal experiment 1997, 3170 m. Plot of the average frequency of flowers versus Julian date (total number of flowers were transformed to a relative frequency from the maximum number of flowers per plot). Julian date 173 corresponds to June 22. Values (± standard error) are based on the mean of five plots.

analysis in the upper site showed that plants in the treatment plots reach their peak of flowering earlier than the controls.

0 0

Post a comment