10.1073/pnas.0704350104

Supporting Information

Files in this Data Supplement:

SI Figure 3
SI Table 1
SI Text




SI Figure 3

Fig. 3. Photographs of experimental layout. (A) Experimental pond array. Drain pipe for implemented drought can be seen on the pond in the center. (B) Mesocosm during imposed drought to show how pond drying was achieved.





Table 1. Comparison of permutation ANOVA (PERMANOVA) and permutation distribution (PERMDIST) among similarity metrics

Functional group and similarity statistic

PERMANOVA

PERMDISP

F

 ratio

P

 value

F

 ratio

P

 value

All Species

     Jaccards

2.13

0.001

34.07

0.001

     Sorensen

2.94

0.001

50.11

0.001

     Chao-Jaccard

5.75

0.001

93.64

0.001

     Chao-Sorensen

6.75

0.001

162.65

0.001

     Bray-Curtis

3.14

0.002

86.93

0.001

Plants

     Jaccards

6.86

0.001

39.49

0.001

     Sorensen

8.33

0.001

71.09

0.001

     Chao-Jaccard

8.33

0.001

428.84

0.001

     Chao-Sorensen

8.33

0.001

479.66

0.001

     Bray-Curtis

4.80

0.002

116.51

0.001

Zooplankton

     Jaccards

1.74

0.002

5.03

0.031

     Sorensen

2.010

0.001

6.36

0.020

     Chao-Jaccard

3.57

0.002

9.99

0.006

     Chao-Sorensen

4.19

0.001

15.41

0.002

     Bray-Curtis

1.87

0.001

11.31

0.006

Herbivores

     Jaccards

1.68

0.046

5.61

0.027

     Sorensen

2.21

0.001

9.67

0.003

     Chao-Jaccard

3.74

0.001

7.65

0.015

     Chao-Sorensen

4.97

0.001

9.56

0.010

     Bray-Curtis

2.54

0.001

4.96

0.039

Predators

     Jaccards

1.31

0.025

11.79

0.003

     Sorensen

1.52

0.023

15.56

0.001

     Chao-Jaccard

1.37

0.031

3.89

0.064

     Chao-Sorensen

1.60

0.031

5.18

0.037

     Bray-Curtis

1.32

0.007

3.17

0.087





SI Text

Comparing compositional similarity through time.

 Over the final 2 years of the experiment, I tested the similarity in community structure in each pond from one year to the next. This allowed me to determine whether pond communities were in flux or whether they had stabilized, such that analyses of the final year would be able to allow me to address my hypotheses. In each year, I calculated the similarity metrics of each pond between the final 2 years of the experiment by using the metrics discussed above. For each treatment, similarity was not statistically different from 1 or from the other treatment (t tests: P > 0.2), indicating that community structure was relatively stable in each pond over the final years of the experiment, and that the treatments did not vary in their degree of change.