Appendix A. Supplementary Figure. Figure A1. Relationship

advertisement
1
Appendix A. Supplementary Figure.
Figure A1. Relationship between algal volume gently packed into a graduated cylinder and
algal wet biomass used to estimate algal addition biomass in the field. Shown here for two
species, A) Ulva sp and B) Ectocarpus. For pooled algae including Ulva lactuca, Ectocarpus,
Gracilaria verrucosa and Bryopsis sp., the relationship between biomass and volume is log(wt)
= 1.07*log(Vol) - 0.87 (r2=0.87, p < 0.001), and the relationship between biomass and percent
cover of 0.5 m2 is log(wt) = 1.07*log(cover) + 0.82 (r2 = 0.94, p < 0.001).
1
Appendix B. Supplementary Statistical Results.
Table A1. Statistical results for tests of cage control treatment versus open treatment (one-way
ANOVA).
Response
MS
df
F
p
Spartina density
28.125
1
0.096
0.768
Error
294.625
6
Proportion flowering
0.007
1
0.184
0.682
Error
0.036
6
Spartina shoot height
2.679
1
0.020
0.895
Error
138.218
5
Above ground biomass
4163.74
1
0.604
0.467
Error
6892.31
6
Below ground biomass
2.565
1
0.136
0.725
Error
18.912
6
Shoot weight
2.639
1
1.672
0.244
Error
1.578
6
2
Tables A2-A6. Statistical results from model comparisons for surveys of benthic primary
producer abundance at restored and natural sites through time. Models were compared using
Akaike Information Criteria adjusted for small sample sizes (AICC) and evidence ratios
calculated as in Burnham and Anderson (2002). Best models have the lowest AICc value, and
the weight of evidence in support of this conclusion is indicated by the evidence ratio with 1
indicating complete support and 0 indicating no support. Models are presented relative to the full
model (Eqn 1), and for each comparison a factor is removed in a logical order.
Table A2. Model results for mean algal abundance (percent cover) from models fit with a
Guassian distribution on arcsine square-root transformed percent cover data. The row in bold
indicates the best model (highest evidence ratio). Factors included in the model are indicated in
the Model column, and RE: indicates random effects.
Log Likelihood
K
AICc
δ
ωi
Ev ratio
369.5
6
-726
35
0
0
366.3
6
-720
35
0
0
Sill, Time and RE: Transect
365.4
5
-720
35
0
0
Sill and RE: Transect
381.6
4
-755
0
1
1
RE: Transect
355.4
3
-705
50
0
0
Model
Full model (Eqn 1): Sill, Time and
RE: Sill | Transect within Site.
Sill, Time and RE: Transect within
Site.
3
Table A3. Macroalgal taxonomic richness model comparison results. Evidence ratios for two
models are small (unconvincing) and similar, so these models are both considered plausible.
Model
Full model (Eqn 1): Sill, Time
Log Likelihood
K
AICc
δ
ωi
Ev ratio
-109.0
6
231
0
1
0.77
-115.5
6
244
13.00
0.00
1.15 x 10-3
-111.7
5
234
3.25
0.20
0.15
-112.3
5
235
4.45
0.11
0.08
-141.6
3
289
45.84
0.00
8.52 x 10-11
and RE: Sill | Transect within
Site.
Sill, Time and RE: Transect
within Site.
Sill, Time and RE: Sill |
Transect.
Sill and RE: Sill | Transect
within Site.
RE: Transect
4
Table A4. Microalgal abundance model comparison results.
Model
Full model (Eqn 1): Sill,
Log Likelihood
K AICc
δ
ωi
Ev ratio
-899.4
6
1811
8.75
0.01
0.01
-901.1
6
1815
12.15
2.23 x 10-3
2.22 x 10-3
-910.1
5
1831
28.00
8.32 x 10-7
8.07 x 10-7
-896.1
5
1803
0
1
0.97
-901.4
4
1811
8.48
0.01
8.52 x 10-11
Time and RE: Sill |
Transect within Site.
Sill, Time and RE:
Transect within Site.
Sill, Time and RE: Sill |
Transect.
Sill and RE: Sill |
Transect within Site.
RE: Transect within Site
5
Table A5. Benthic algal primary production (macroalgae + microalgae abundance).
Model
Full model (Eqn 1): Sill,
Log Likelihood
K
AICc
δ
ωi
Ev ratio
-975.0
6
1963
0
1
0.89
-979.6
6
1972
9.2
0.01
8.98 x 10-3
-979.2
5
1969
6.25
0.04
0.04
-978.8
5
1968
5.44
0.7
0.06
-999.9
4
2008
45.5
Time and RE: Sill |
Transect within Site.
Sill, Time and RE:
Transect within Site.
Sill, Time and RE: Sill |
Transect.
Sill and RE: Sill | Transect
within Site.
RE: Transect within Site
1.30 x 10-10 1.16 x 10-10
6
Table A6. Marsh cordgrass Spartina alterniflora abundance. Evidence ratios for two models are
small (unconvincing) and similar, so these models are both considered plausible.
Model
Log Likelihood
K AICc
δ
ωi
Ev ratio
-15.97
6
44.44
24.4
4.96 x 10-6
2.65 x 10-6
-17.04
6
46.58 26.57
1.70 x 10-6
9.12 x 10-7
-13.73
5
37.82
17.8
1.36 x 10-4
7.30 x 10-5
Sill and RE: Sill |Transect.
-7.47
4
23.18
3.16
0.21
0.11
RE: Sill | Transect
-5.89
4
20.02
0
1
0.54
-7.35
3
20.82
0.82
0.66
0.35
Full model (Eqn 1): Sill,
Time and RE: Sill |
Transect within Site.
Sill, Time and RE:
Transect within Site.
Sill, Time and RE: Sill |
Transect.
within Site.
RE: Transect within Site
7
Download